social crowdfunding: individual- and project-related ... · kind of fundraising, websites with...

Social crowdfunding: individual- and project-related determinants of success Empirical investigation on Kickstarter.com Master Thesis Erasmus School of Economics Economics and Business, Marketing

Name: Delyan Peyankov, 369184

Supervisor: dr. Hyoryung Nam

“There’s just something magical about Kickstarter... You immediately feel like you’re part of a larger club of art-supporting fanatics.” Amanda Palmer, who rallied 25,000 supporters for her album

http://www.kickstarter.com/projects/amandapalmer/amanda-palmer-the-new-record-art-book-and-tour

1 | P a g e

2 | P a g e

Preface

This Master Thesis is part of my graduation program at Erasmus School of Economics on

Economics and Business, specialization Marketing. I would like to express my gratitude towards

my thesis supervisor dr. Hyoryung Nam for the opportunity to work on such an interesting and

challenging topic. Throughout the past few months I was given her constant support and

comprehensive feedback, without which I would not be able to progress on this Master Thesis. I

would further like to thank Kickstarter for not suspending the necessary data collection cycle. It

was long and technically demanding process during which I needed their collaboration. Without

this support, the current Master Thesis would not contain such a broad sample.

Last but not least, I would like to show my appreciation to the Marketing department for the

organization of the curriculum in a way that I had all the necessary academic skills to facilitate

the writing of this Master Thesis.

3 | P a g e

Abstract

The current Master Thesis aims to analyze individual- and project-based factors of crowdfunding

success. This revolutionary new fundraising form redefines the meaning of entrepreneurship

and expands the boundaries of funding opportunities. Due to the relative novelty of the

phenomenon academic papers on that topic are still limited. Nevertheless, the present Master

Thesis adopts literature from the fields of marketing, economics and social psychology to create

unique viewpoint on crowdfunding. By doing so, the research contributes to the existing

academic lit by providing new insights with potentially high managerial impact. The conceptual

model distinguishes between 6 individual- and project-related factors and 7 hypotheses are

developed. In order to test them a sample of 6294 projects from the biggest crowdfunding

platform Kickstarter are collected and analyzed. In total 6 direct effects are identified. Next to

this one partial and full mediation, as well as 3 moderation effects are identified. Based on these

results the paper builds elaborative recommendations aiming at entrepreneurs who consider

the use of crowdfunding for funding their projects.

4 | P a g e

Contents

1 Introduction ........................................................................................................................................... 6

1.1 The Emergence of Crowdfunding .................................................................................................. 6

1.2 Definition of Crowdfunding ........................................................................................................... 7

1.3 Relevance and Research Objectives .............................................................................................. 8

1.3.1 Academic Relevance .............................................................................................................. 8

1.3.2 Managerial Relevance ........................................................................................................... 9

1.4 Thesis Outline ................................................................................................................................ 9

2 Theoretical Background ...................................................................................................................... 10

2.1 From Crowdsourcing to Crowdfunding ....................................................................................... 10

2.2 Crowdfunding Literature ............................................................................................................. 12

2.2.1 Customers as Investors........................................................................................................ 12

2.2.2 Classification of the Crowdfunding System ......................................................................... 13

2.2.3 Inter-Personal Relations in Crowdfunding .......................................................................... 14

2.2.4 Helping Behavior ................................................................................................................. 15

3 Conceptual Model ............................................................................................................................... 16

3.1 The role of the Social Ties............................................................................................................ 17

3.1.1 Information Asymmetry ...................................................................................................... 17

3.1.2 Social Ties ............................................................................................................................ 17

3.2 Reputation ................................................................................................................................... 18

3.3 Online community - Membership and Involvement ................................................................... 19

3.3.1 Crowdfunding Platforms seen as Online Communities ....................................................... 19

3.3.2 Membership and Involvement ............................................................................................ 20

3.4 Credibility .................................................................................................................................... 21

3.5 Request Size ................................................................................................................................ 22

4 Data ..................................................................................................................................................... 23

4.1 Kickstarter.com as Platform of Analysis ...................................................................................... 23

4.2 Data Collection Design ................................................................................................................ 24

4.3 Project Landing Page – 2nd layer ................................................................................................ 26

4.4 Profile page – 3rd layer ............................................................................................................... 27

5 | P a g e

5 Measures ............................................................................................................................................. 28

5.1 Project Success ............................................................................................................................ 29

5.2 Direct Social Ties .......................................................................................................................... 30

5.3 Reputation ................................................................................................................................... 30

5.4 Community Membership ............................................................................................................ 31

5.5 Community Involvement ............................................................................................................. 31

5.6 Credibility .................................................................................................................................... 31

5.7 Request Size ................................................................................................................................ 32

5.8 Dynamics of the Measures .......................................................................................................... 32

6 Results ................................................................................................................................................. 33

6.1 Descriptive Statistics.................................................................................................................... 33

6.2 Correlation Coefficients ............................................................................................................... 34

6.3 Regression Models ...................................................................................................................... 35

6.3.1 Validity and Assumptions .................................................................................................... 35

6.3.2 Multiple Models .................................................................................................................. 38

7 Discussion on the Results .................................................................................................................... 41

7.1 Direct Social Ties .......................................................................................................................... 41

7.2 Reputation ................................................................................................................................... 41

7.3 Community Membership and Involvement ................................................................................ 42

7.3.1 Community membership ..................................................................................................... 42

7.3.2 Community involvement ..................................................................................................... 43

7.4 Credibility .................................................................................................................................... 44

7.5 Request Size ................................................................................................................................ 45

7.6 Project Category .......................................................................................................................... 46

8 Conclusion ........................................................................................................................................... 46

8.1 Research Relevance ..................................................................................................................... 46

8.2 Summary of the Findings ............................................................................................................. 47

8.3 Limitations and Future Research ................................................................................................. 48

8.4 Research Contributions ............................................................................................................... 49

8.5 Managerial Implications .............................................................................................................. 50

9 References: .......................................................................................................................................... 52

10 Appendix:......................................................................................................................................... 57

6 | P a g e

1 Introduction

1.1 The Emergence of Crowdfunding

The past few decades have witnessed an increase in enthusiasm for entrepreneurship and

entrepreneurs. Next to practitioners, academics have matched the same enthusiasm. This

broad attention can be justified though since new business creation is the major source of

economic growth and agility, new job creation and technological innovation (Birley 1987,

Reynolds 1987).

Being an entrepreneur and initiator of entrepreneurship ventures is often risky occupation

accompanied with uncertainties. The first problem entrepreneurs face at the very beginning

of their initiative is the accessibility to capital. New ventures face difficulties attracting

external investments at their initial stage from equity capital and bank loans (Hughes

2009). While venture capitalists and business angels fill the gap for high amount initiatives,

smaller amount ventures mostly rely on internal financial power or support f rom friends and

family (Belleflamme et al. 2010). Often known as bootstrapping techniques their use aim to

mitigate the entrepreneurs’ financial constraints (Bhide 1992, Ebben 2006). However, due to

the huge financial risk, funding limitations and instability many entrepreneurs are often forced

to give up.

In the recent years, some entrepreneurs have started to seek financial support from the public

or the ‘crowd’ via internet, instead of approaching banks, venture capitalists or business angels

(Kleemann et al. 2008, Belleflamme et al. 2010). This funding technique is referred to as

“crowdfunding”. Crowdfunding is becoming possible due to the emergence of ‘Web 2.0’, a

technology beyond the static webpages, which allows extended usage of internet

applications and two-way communication through its interactive and collaborative structures

(O'Reilly 2005, Kleemann et al. 2008).

7 | P a g e

The crowdfunding ‘phenomenon’ comes naturally, since from 1990s, the World Wide Web

plays an important role in shaping and enhancing the relationships between consumers and

companies. C rowdfunding is just a reflection of the immense possibilities of ‘Web 2.0’

technology. Nevertheless, the roots of crowdfunding and can be found in the crowdsourcing.

Crowdsourcing uses the ‘crowd’ as a generator of ideas, solutions and feedback for certain

company’s problems or processes (Belleflamme et al. 2010). The use of user-generated content

in blogs, forums, social networks (Twitter, Facebook, Pinterest, etc.), special crowdsourcing

platforms (Innocentive, iStockphoto, Dell IdeaStorm, etc.), or open-source software (Linux,

Mozilla Firefox, WordPress, etc.) can be seen as crowdsourcing initiative. In a broad sense

crowdfunding can be seen as an extension of the crowdsourcing. In fact some crowdfunding

projects can indeed encourage participation of the supporters for the product development,

innovation and decision-making processes, which are typically crowdsourcing initiatives

(Belleflamme et al. 2010).

1.2 Definition of Crowdfunding

Belleflamme et al. (2010) define crowdfunding as “an open call, essentially through the

Internet, for the provision of financial resources either in form of donation or in exchange for

some form of reward and/or voting rights in order to support initiatives for specific purposes”.

Instead of seeking funds from a relatively small group of sophisticated investors (banks,

venture capitalists, etc.), entrepreneurial projects can obtain the necessary funds from the

‘crowd’, using internet-based platforms where individuals can support the project with

relatively small amounts of money. In exchange of the pledge, the backer usually receives a

“reward” which varies from a ‘thank you message’ to the actual product, a share or premium

content depending on his pledge. Initially crowdfunding has been used in the entertainment

music and film industry (Belleflamme et al. 2010). The first known crowdfunding platform is

the US based company ArtistShare in 2001, later on followed by PledgeMusic and Sellaband in

2006. However, not long after that many entrepreneurs saw the opportunity into that new

kind of fundraising, websites with variety of categories like Indiegogo (2008) and Kickstarter

(2009) became common (Zouhali 2011 ).

8 | P a g e

While crowdfunding has been primarily used for small entrepreneurship ventures this is not

always the case. Currently, there are number of projects rising hundreds or thousands of

dollars each. An example of such successful initiative is the popular project from 2012 in

Kickstarter – ‘Pebble: E-Paper Watch for iPhone and Android’ which is pledged for 10.3

million, supported by almost 70 000 backers (Smith 2013). The success of that initiative

demonstrates that crowdfunding is potentially a fundraising mean not only for small

projects, but also for ambitious startups, typically funded by venture capitalists or business

angels (Belleflamme et al. 2010). There are over 500 crowdfunding platforms by the end of

2012 and their number is still growing. A research from ‘Massolution’ predicted the

crowdfunding industry (equity, donation, lending and reward) in 2012 is estimated at $2.8

billion, and the number are exponentially growing (Caldbeck 2012).

1.3 Relevance and Research Objectives

1.3.1 Academic Relevance

That apparent interest in the crowdfunding phenomenon gets the attention of the press and the

media (e.g. ‘Inc.’,’ Forbes’, ‘Bloomberg’, bloggers and social media) which try to follow with the

emerging technology and business trends. In contrast, the academic research on the topic is

considerably limited and so far published works in recognized journals are almost absent.

Nevertheless, the growing interest from academics indicates that the crowdfunding is becoming

popular among scholars. Recently this area received some attention from academics (e.g. Bayus

(2013), Ordanini et al. (2011), Belleflamme et al. (2010), Agrawal et al. (2011) and others).

However, still the understanding on the underlying mechanisms is quite fragmented and it

remains rather unknown what are the driving forces behind the crowdfunding success. The goal

of the current thesis is to contribute to such limited understanding by empirical investigation of

the main individual- and project-related determinants from the crowdfunding platform

Kickstarter.

9 | P a g e

This paper adopts various academic articles from marketing, social and behavior literature to

construct comprehensive quantitative research. In a way the current paper adds up to the

research from Evers (2012) and expands his findings by explaining the aspects of the social

network connections, community reputation, membership and involvement in the context of

online social crowdfunding. To the best of our knowledge no research so far has examined these

aspects of the crowdfunding, so it is believed that the current paper will reveal new insights

with academic and managerial relevance.

1.3.2 Managerial Relevance

From a managerial perspective the research is relevant as well. It is apparent that the

emerging crowdfunding trend grows in users and popularity. Entrepreneurs seek to fund their

projects and often the only viable option is to tap into the crowd. Furthermore, the recent

financial crisis negatively affected the private sector and external financial sources became

scarce (Dell'Ariccia 2008). This trend places the emphasis on that new online mean of funding.

New insights in this phenomenon will help practitioners to make better decisions on their

projects. Results of the current research reveal important determinants on individual- and

project-levels. For example, the importance of the community involvement and the social buzz

influence are factors to take into consideration before setting a project. Additionally, the

effects of request size and the variety of funding tiers are findings with highly practical

implications. Entrepreneur setting an online crowdfunding project may consider the insights

provided from the current paper before initiating a project.

1.4 Thesis Outline

Firstly, existing theoretical background on the crowdfunding phenomenon is presented. Existing

academic papers on the topic are limited and most of them are briefly discussed. Next, the

development of the conceptual model is presented along with the hypotheses. Relevant

marketing and behavior literature is borrowed for construction points. Following is the data

collection design and a short description of some key elements of the website. Measures used

for quantifying the conceptual model and testing the hypotheses are explained in the next

10 | P a g e

chapters. Discussion on the results along with managerial implications findings follow.

Conclusion is provided at the end, where academic contributions and managerial applications

are discussed. Lastly limitations and directions for future research are pointed out.

2 Theoretical Background

2.1 From Crowdsourcing to Crowdfunding

Kleemann et al. (2008) asserts that “crowdsourcing takes place when a profit oriented firm

outsources specific tasks essential for the making or sale of its product to the general public (the

crowd) in the form of an open call over the internet”. The idea is that the contribution of the

individuals is ether for free or significantly less than its actual worth (Kleemann et al. 2008).

Firms engage creative individuals, highly skilled professionals, or the collective opinion of a

group of individuals (“Wisdom of the crowd”1) to generate value. Crowdsourcing tasks include

technical solutions, product or graphic design, advertising, quality control, etc.

The functional differentiation of society into "producers" and "consumers" is an old-fashioned

concept of the industrial society (Kleemann et al. 2008). Nowadays, consumers have stopped to

be entirely takers of goods and services, but rather they start to take part into the firms’

production and delivery processes. This describes consumers more like co-workers, taking

specific production processes under the control of the enterprise (Kleemann et al. 2008). The

crowdsourcing phenomenon is accompanied with the emergence of that specific type of

consumer - the "working consumer” (Kleemann et al. 2008). Howe (2006) compares

crowdsourcing to outsourcing, asserting that the new pool of cheap labor are people taking

part into the content creation, innovation and R&D processes. Whereas some outsource their

processes abroad looking for cheap labor, the development of “Web 2.0” platforms made

possible for others to tap into the unlimited supply of potential workers (Howe 2006).

1 See James Surowiecki – “The Wisdom of Crowds” (2005), ISBN 978-0-385-50386-0

11 | P a g e

In a broad sense the concept of crowdfunding can be seen as integral part of the crowdsourcing,

using the “crowd" as a source of ideas, feedback and solutions for developing business activities

(Belleflamme et al. 2010). The crowdsourcing definition can provide a key understanding why

crowdfunding is embedded into the crowdsourcing (Belleflamme et al. 2010). In crowdfunding

initiatives the consumers can volunteer to provide help into the development of a product or

the support of a cause and that input is in the form of financial support. From this perspective,

crowdfunding is a subset of crowdsourcing, since the latter includes also financial help

(Belleflamme et al. 2010). Similarly to crowdsourcing, crowdfunding uses online social

communities or networks to finance the funding initiatives (Kuile 2011). Hence, that social

element in the crowdfunding is the main pillar in the crowdfunding existence and development.

Crowdfunding has become popular in the recent 3-4 years, compared to the crowdsourcing

which dates back from the beginning of the century. The emergence of big crowdfunding

platforms such as Sellaband (2006), Indiegogo.com (2008), and Kickstarter.com (2009) has

become the starting point of that revolutionary new funding mean. The comparison of Google

searches for the two terms has been shown on the figure below in order to contrast the

emergence of the two concepts.

Figure 1 - Google search results

12 | P a g e

2.2 Crowdfunding Literature

As crowdfunding is relatively new phenomenon it is not a surprise that published academic

literature on that topic is considerably limited. So far very few scholars dealt exclusively with the

topic of crowdfunding (Kappel 2009, Belleflamme et al. 2010, Schwienbacher 2010, Agrawal et

al. 2011, Belleflamme et al. 2011, Ordanini et al. 2011, Bayus 2013). Extensive literature exists

on the topic of crowdsourcing (Kleemann et al. 2008, Malone et al. 2010, Chong 2012, Poetz

2012, Stieger et al. 2012). However, most of the crowdsourcing literature does provide insights

for understanding the crowd used as a source of funding.

Another type of literature, related to the helping behavior (Bendapudi et al. 1996),

entrepreneurship (Shane 2002), social and group behavior (Dholakia et al. 2004) is utilized for

the conceptual model of that research. However, it is hard to fully translate all of the borrowed

literature to “Web 2.0” context, inherent to the crowdfunding. This limitation may potentially

constrain the theoretical model.

In the following paragraphs significant academic works dealt exclusively with the crowdfunding

matter are briefly discussed.

2.2.1 Customers as Investors

Ordanini et al. (2011) set an in-depth qualitative analysis among a set of diverse in nature

crowdfunding operations - SellaBand, Trampoline and Kapipal. The paper addresses

fundamental questions of why and how consumers turn into crowdfunding participants and the

service providers to set up a crowd-funding initiative. The research clarifies that the project

involvement, the social engagement and the monetary return are among the main drivers for

consumers to become investors. However, these factors heavily depend on the crowdfunding

platform. The diverse set of platforms examined in that research tend to catch the differences

among them. Despite the differences common for all is that consumers-investors like engaging

in innovative behavior (Ordanini et al. 2011). Beyond the mere content of the crowdfunding

13 | P a g e

initiative, the participants are also attracted by the innovative way the technology platform is

used, especially in the context of social networking (Ordanini et al. 2011). Indeed important

point of interest for the current paper is the social networking aspect of crowdfunding, which

does not get the necessary attention by academics.

Ordanini et al. (2011) study contributes also to the crowdfunding literature describing the

project’s diffusion curve. The authors identify 3 phases – rapid accumulation in the beginning

mainly due to the investments of the users who are directly connected to the project (e.g.

friends, families, colleagues, etc.), slowdown in the middle described as “getting the crowd”

phase where word-of-mouth determine whether the project will turn into the third phase,

described as an engagement phase, or the phase where the funding is done mainly by people

who are interested in the project.

However, despite the obvious contributions to the crowdfunding literature the study’s

qualitative approach constrain to a certain extent the scope of the paper. Further quantitative

research is necessary to match the same findings with regards to project involvement, social

engagement and the monetary return as main drivers for consumers to become investors.

2.2.2 Classification of the Crowdfunding System

Belleflamme et al. (2010) conduct an empirical study from industrial organization point of view

on different types of rewards and rights, as well as the magnitude of the financial contributions

generated through crowdfunding. The research is based on 51 crowdfunding ventures. The

paper distinguishes between 3 different forms of investment: donations, active investments,

and passive investments. The study discovered that non-profit initiatives are significantly more

likely to reach their funding goal than for-profit ones and project based initiatives. Belleflamme

et al. (2010) assert that crowdfunding initiatives have implications beyond the financial sphere

(e.g. having entirely functional goal to fund the project), it also draws the public attention and

receives customer response by using the crowdfunding platform as a two-way communication

14 | P a g e

channel. Hence, crowdfunding may also be regarded as a powerful promotional tool and user-

based innovation tool.

Schwienbacher (2010) provide extensive analysis of crowdfunding practices based on a study

case from a French startup. The paper emphasizes on the practical implications by explaining

when small entrepreneurial ventures can successfully use the crowdfunding approach, instead

of other fundraising techniques. Schwienbacher (2010) defines crowdfunding in very simple

terms: “Financing of a project or a venture by a group of individuals instead of professional

parties (like, for instance, banks, venture capitalists or business angels)”. By any means rising

money is the most important reason behind setting up a crowdfunding initiative. However,

Schwienbacher (2010) recognizes two more important reasons. Firstly, creating public attention

and secondly testing the market potential before bringing the product to the market. Similar to

Belleflamme et al. (2010) findings, crowdfunding goes beyond simply rising money for initiating

a project, but may be used for effective product management. In that respect the research

recognizes the importance of the efficient communication and networking. In order to reach

skilled motivated supporters, it is crucial to reach as many people as possible in the first place

(Schwienbacher 2010). That simply means that the pool of potential funders is increasing as the

network of people exposed to any form of communication about the project is growing. In that

context the effective communication through Web 2.0 such as social networks or even the use

of offline communications is critical for creating the buzz (Schwienbacher 2010).

2.2.3 Inter-Personal Relations in Crowdfunding

Agrawal et al. (2011) explore the role of the geographical distance in the early stage

crowdfunding entrepreneurial projects. They set a quantitative research on investments among

34 entrepreneurs in the music industry. One of the main results consistent with the

entrepreneur literature is that the local circles (likely friends and family) are among the first and

most important source of capital in the first stages of the entrepreneur’s venture and the same

are less influenced by the other investors. These findings are also in line with the crowdfunding

diffusion curve set by Ordanini et al. (2011). Agrawal et al. (2011) explain that finding with the

15 | P a g e

information asymmetry. Similar findings are discussed by Shane (2002) in the context of venture

funding, which is adopted in the conceptual model of the current thesis. However, consistent

with prior research Agrawal et al. (2011) results suggest that even though online crowdfunding

seems to eliminate most distance-related economic frictions (e.g. providing input, gathering

information), it does not eliminate social-related frictions.

2.2.4 Helping Behavior

Sources from other literature are also related to the crowdfunding phenomenon. Number of

marketing literature exists on the topic of helping behavior. However, the majority of them is

focused on specific aspects of the helping behavior. For example a great amount of literature

deals with the type of request (Burnkrant 1982, LaTour 1989, Bagozzi 1994) and its size (Reingen

1978, Brockner et al. 1984, Schibrowsky 1995). Bendapudi et al. (1996) utilize that to create a

theoretical framework for understanding the drivers affecting the helping. The framework

provides a broad overview beyond the mere main effects and provides investigation over the

interaction effects of donor and charity variables. The same is adopted in the conceptual model

of the current study.

Evers (2012) utilizes the framework from Bendapudi et al. (1996) and adopts it in the context of

crowdfunding. The research is based on 8807 projects from the popular crowdfunding platform

IndieGoGo. Totally 8 drivers of success have been identified - cause of need, image, picture

appeal, social comparisons, labeling, perspective advocated, decisional control and request size.

Evers (2012) provides extensive research using actual data from the website in order to present

the most important factors of success in any crowdfunding initiative on IndieGoGo. It is worth

mentioning that his research is the first that presents the complex issue in such extensive way

and with that it has significant managerial implications. As noted before, the current research

expands on his research and further explains additional aspects of crowdfunding, potentially

having a high managerial impact.

16 | P a g e

3 Conceptual Model

To develop the conceptual model, theories from marketing, economics and psychology are

borrowed. They are utilized accordingly in the context of social crowdfunding and seven

hypotheses are constructed. The relevant literature along with the hypotheses are presented in

the current chapter.

The conceptual model discriminates between individual-focused factors (network ties,

community reputation, membership and involvement) and project-focused factors (image

credibility and request size). The project outcome is the main dependent variable in the model,

which is measured by the actual project success rate. Along with the direct effects one

mediation effect is tested. Visual reference to the model is presented below:

Figure 2 - Conceptual Model

17 | P a g e

3.1 The role of the Social Ties

3.1.1 Information Asymmetry

In the context of venture funding the information asymmetry can be the biggest setback

between potential investor and the entrepreneur. The information asymmetry exists when “one

of the party has less perfect information than the other party” (Johnsen 2010). The information

asymmetry assumes that the relationship between the entrepreneur and the potential investors

is not flat, so as the dissemination of information (Shane 2002). For instance innovative

technology start-up seeking for external finance is often not willing to disclose all the know-how

to potential investors, since doing so may lead to a leak of important information, likely to be

used by competitors. In that case the investor has to decide on the funding not in the presence

of perfect information. Theoretically, the information asymmetry described above could

discourage potential investors and make it hard for the entrepreneur to obtain external funding

(Venkataraman 1997, Shane 2002). Despite that claim, however, we do not observe failure in

the venture finance market (Shane 2002).

3.1.2 Social Ties

The study from Shane (2002) asserts that social ties are the main mechanism for overcoming

the information asymmetry in venture entrepreneurship. Direct social tie is “a personal

relationship between a decision maker and the party about whom the decision is being made”

(Larson 1992). In the context of online social networks a person is part of a user’s 1st level

network if the same has a direct connection with the former (Trusov et al. 2010). That direct tie

or the entrepreneur-investor dyad is a result of prior activity (Shane 2002). That prior

relationship is the reason for funding decisions to be affected through social ties. Information

transfer through social connections and social obligations between both parties are the two

main mechanisms through which social ties affect venture funding decisions (Shane 2002). By

possessing a network of direct ties the entrepreneur can share information and step into a

transactional relationship where both parties are motivated to maintain relationship in a

18 | P a g e

trustful manner generating a sense of obligation (Shane 2002). That potentially overcomes the

information asymmetry problem as information is shared without any risk for the entrepreneur

or the investor. Hence previously established direct tie between the investor and the

entrepreneur makes the investment more likely to take place, than without such a link (Shane

2002). However, the impact of direct social ties on investment decisions has not been discussed

in online setting. Despite that for the current research existing theory is adopted and it has been

assumed it is relevant for the online setting. That assumption might be misleading, because the

information asymmetry in the online setting is limited, compared to the offline setting.

However, considering the shortage of literature and in line with the exploratory nature of the

research it is hypothesized the following:

H1: The number of previously established direct social ties (1st level network) of the

entrepreneur favorably affects the project success.

3.2 Reputation

Shane (2002) discusses entrepreneur’s reputation alongside social relationships. Reputation is

defined as information about individual’s past performance (Podolny 1994). Reputation

provides information to the investors about the entrepreneur’s skills, expertise and possessed

abilities to bring the project to its launch (Shane 2002). Once the information about the

entrepreneur’s past performance is known then his abilities are beyond dispute. Therefore

positive reputation should attract investment interest, since the entrepreneur has already

proved his successfulness (Shane 2002). That suggests that investors should be more likely to

fund the entrepreneurship project once the entrepreneur has positive reputation about his past

performance (Shane 2002). Hence it is hypothesized the following:

H2a: Previously established positive reputation of the entrepreneur favorably affects the

project success, whereas negative reputation withdraws investment interest and unfavorably

affects the project success.

Reputation assumes that information about entrepreneur’s performance is publicly available

(Shane 2002). Once the information about the entrepreneur’s reputation is publicly available

19 | P a g e

the efforts of investors to obtain private information from social network ties would become

less relevant (Shane 2002). In the context of social crowdfunding communities this behavior

suggests investment decisions based on possessed entrepreneur’s crowdfunding reputation

rather than based on possessed investor social networking. To the extent that social network

ties are used for collecting information, reputation should mediate the effect of social ties on

the funding decision (Scott Shane 2002). Hence it is hypothesized:

H2b: The role of the previously established reputation of the entrepreneur mediates the

effect of direct network ties on the project success.

The above thoughts assume that social ties are a mean to gather information (Shane 2002). That

is a twofold process, from the potential investor and from the entrepreneur. Based on

entrepreneur’s reputation mediation should exists only if the investor is the one who seeks for

information. However, it is often the entrepreneur who distributes the information and the

investor passively adopts the information. That behavior is especially common across social

networks. In that case the reputation should act only as a signaling variable, without mediating

the role of the social ties. Based on this we can expect partial mediation of the reputation over

the effect of direct ties on the project success rate.

3.3 Online community - Membership and Involvement

3.3.1 Crowdfunding Platforms seen as Online Communities

Virtual communities are defined to be digital environment where users organize, communicate

and support each other. It is a special type of social entity mediated by the digital environment

(Rheingold 1993, Bagozzi 2002). The shared goal of the online communities can be either

functional (for e.g. exchange of information or support of the community members) or/and

hedonic (e.g. user experiences and interactions between its members) (Bagozzi 2002). However,

in both cases community is the essential reference group for its users (Bagozzi 2002). A

resemblance is found between the above mentioned characteristics and those of the social

crowdfunding platforms. First, the communication between the members of the community is

entirely mediated by the online setting. Next, there is a clearly defined functional goal within

20 | P a g e

the crowdfunding communities, such as the mutual benefit for both the project initiator as he

generates financial resources and the supporters as they receive reward in return of their

pledge. Finally, hedonic motives are also drivers of participation as for instance the joy of

interaction with the members or the shared interest in certain crowdfunding projects are

among the many examples for driving behavior. As a common rule online social communities

depend on users’ voluntary participation, commitment, and contributions (Bagozzi 2002).

3.3.2 Membership and Involvement

For conceptualizing the role of the community membership and involvement the broader

concepts of social identity and in-group favoritism are adopted.

Social identity represents the main aspects of the individual’s identification with the community

or the group in a way that one view him- or herself as a member of the group or the community

to which he/she belongs (Dholakia et al. 2004). Social identity is part of the individual's self-

concept derived from perceived membership in a relevant social group (Turner 1986). This

assumes that membership is the main construct of social identity. Further social identity enforce

group-oriented behavior, that leads to stronger online community participations, driven by the

in-group favoritism (Ellemers 1999, Dholakia et al. 2004). Hence, community participations or

otherwise involvement is affected by in-group favoritism, determined by the social identity. In

the context of social crowdfunding this means that members of the community would be more

inclined to support community activities and in return be an object of support from other

community members.

In-group favoritism refers to a pattern of favoring members of one’s in-group over out-group

members (Dasgupta 2004). It is worth mentioning that in-group favoritism may exist even

within members of the same online community. This is deemed to be possible in the context of

online communities, since membership and involvement in the online setting are usually not

strictly defined as they are in offline setting. For example new members are free to join and quit

at any time and community members may merely exist, without being active whatsoever. In this

21 | P a g e

case such members are not perceived as healthy members of the community, but rather seen as

outsiders.

The concept of social identity discriminates between two constructs that affect and in return are

being affected by in-group behavior - perceived membership and involvement and participation

within the online community. Similarly, the current study also discriminates between

membership and involvement. Community membership represents the extent to which a user is

identified with the online community as a function of his membership (e.g. being a member of

the community and sharing the same social norms with the rest of the members). It is worth

mentioning that community membership differs from community reputation or involvement, as

the former represents the passive engagement of the member and in return adopts the

community’s social norms which do not assume active behavior. Community involvement on

the other hand represents the extent to which the user actively participates in online

community activities (e.g. support members of the crowdfunding community, by being

financially or non-financially involved – by commenting, backing, sharing content, etc.). It is

important to note that due to their virtual existence membership and involvement in online

communities are matters of degree which vary across members (Bagozzi 2002). Some users

might be actively engaged in the online community (e.g. supporting, commenting, sharing),

whereas others may merely be present without being involved.

Based on the above theoretical reasoning it is hypothesized the following:

H3: Higher entrepreneur’s membership within the online community favorably affects the

project success.

H4: Higher entrepreneur’s social involvement within the online community favorably affects the

project success.

3.4 Credibility

As discussed before Bendapudi et al. (1996) create a framework explaining the helping behavior.

The paper identifies number of factors affecting the helping behavior – antecedents or charity

22 | P a g e

controlled factors (source variables, message variables and request variables) and moderators

(donor variables and non-donor variables). These factors lead to helping behavior (serious help,

token help or no help), which on the other hand leads to a number of consequences for the

community, for the donor, the charity or the beneficiary (Bendapudi et al. 1996). It is arbitrary

whether that framework is fully applicable in the context of crowdfunding for a number of

reasons (see section 2.3.4). However, part of it is adopted to explain important factors in the

crowdfunding project, such as the credibility of the source and request variables which we

consider having an important role for the project outcome. It is worth mentioning that these

factors are project-related, whereas the factors discussed so far were individual or creator-

focused.

As a source variable, the charity image is important determinant of the helping behavior, since

the message is usually a function of the charity’s image (Bendapudi et al. 1996). Both, the

message and the source image are antecedents of the helping behavior and may be affected by

a number of moderating factors, as explained above (Bendapudi et al. 1996). However, since the

message is a function of the source image, the last is a critical element of the charity outcome

and determines whether contribution is considered (Bendapudi et al. 1996). In order for

potential supporters to consider contribution, they must first believe in the message, which is a

function of the source image (Bendapudi et al. 1996). In the context of social crowdfunding the

source image is the overall image of the entrepreneur or the project creator, created by

external or internal cues. Social literature asserts that the message is more likely to be accepted

when the source has credible and familiar image (Kelman 1961). Therefore it is hypothesized:

H5: Credible image of the entrepreneur is more likely to be accepted by potential

investors and favorably affects the project success.

3.5 Request Size

Bendapudi et al. (1996) consider request drivers and particularly the size of the request as

antecedent which also affect the helping behavior, thus the outcome of the fundraising project.

The relationship between the requested size and the project outcome is not always clear.

23 | P a g e

Higher request size may be perceived as unreasonable, whereas lower request size may lead to

rejection because the potential supporter sees his contribution as insignificant. This assumes

that the size of the request should be significant, yet reasonable. In this case the ideal approach

that would maximize the funding outcome is to set higher request size to traditionally high

supporters and lower request size to traditionally low supporters (Schibrowsky 1995). However,

very often previous supporter’s behavior is not observed by the entrepreneur and that

maximization strategy is hard to be applied. Without such information, literature studies

observe the answer is in line with the notation that the request size for fundraising projects is

usually a subject to price effects. Schibrowsky (1995) asserts that the response rate is usually

inversely related to the size of the request. Hence it is hypothesized the following:

H6: Higher request size should lead to lower response rate which unfavorably affects the

project success.

4 Data

In order to test the hypotheses actual data is collected from the crowdfunding platform

Kickstarter.com. In the current section the platform of the analysis, the data collection design

and the website specifications relevant for the study are briefly described.

4.1 Kickstarter.com as Platform of Analysis

The data of the current thesis is entirely based on the crowdfunding platform Kickstarter.com.

We will briefly introduce the website, the officially published statistics and the general

characteristics of it.

Kickstarter was launched in April, 2009, first known as ‘KickStartr’. Soon after that the website

was awarded by ‘New York Times’ and ‘Time’ as one of the ‘Best Inventions of 2010’ and ‘Best

Websites of 2011’ (Snyder 2010, McCracken 2011). As for the official Kickstarter statistics,

102,372 projects were launched till July 2013. From all 45,037 were successful and pledged for

24 | P a g e

$606 million (Kickstarter Statistics Jul 14 2013). This makes the website the biggest

crowdfunding platform in the world at the moment (Blanca 2012).

Kickstarter uses ‘All or Nothing’ model, which means that if the project does not reach its goal

within the timeframe, it will not be funded and the pledges collected will be returned to the

backers. Kickstarter charges 5% of the funds raised. Amazon Payments (the official payment

mean of Kickstarter) charges additional 3–5% (Kickstarter Webpage 2013). Unlike many other

platforms, Kickstarter does not claim ownership over the projects. Kickstarter projects are

classified in 13 categories namely: ‘Film & Video’, ‘Music’, ‘Publishing’, ‘Art’, ‘Games’, ‘Design’,

‘Fashion’, ‘Food’, ‘Photography’, ‘Theater’, ‘Technology’, ‘Comics’, and ‘Dance’ and 50 sub-

categories. The most initiated category is ‘Film & Video’, but it is not the most successful one,

nor the most pledged one (Kickstarter Statistics Jul 14 2013). Please refer to the table below

with the most recent Kickstarter statistics from July 14th 2013.

Table 1 - Kickstarter Statistics July 14th 2013

4.2 Data Collection Design

The data was extracted from actual Kickstarter projects, all of which have finished by the time

they were included in the sample. Visual Web Ripper 2.12 data-extraction software was used for

that purpose. After the collection plan was designed, the software automatically scans through

Project Category Initiated Projects Successfully Funded Projects

Film & Video 26842 39.92%

Music 22697 54.88%

Publishing 12040 32.22%

Art 9252 48.87%

Games 5627 34.71%

Theater 4713 64.42%

Design 4290 38.72%

Food 3746 41.06%

Fashion 3520 28.89%

Photography 3221 36.57%

Comics 2649 48.66%

Technology 2410 34.02%

Dance 1365 71.36%

Total/Average 102372 43.99%

25 | P a g e

2 webpages - Kicktraq.com and Kickstarter.com in order to collect various website content

structures and to organize them into usable spreadsheet format.

Kickstarter.com has the policy not to disclose unsuccessful projects and not to publicly share any

kind of information related to them. Therefore within the Kickstarter.com webpage the user

cannot navigate to unsuccessful or canceled projects. Nonetheless, the same are accessible if

the exact link or the name of the project is entered into the search bar. The limitation was a

huge setback at the initial stage of the research, since it seemed impossible to obtain

information from unsuccessful projects, because they were virtually inaccessible through the

navigation bars in Kickstarter. This potentially means unsuccessful projects would not be

collected, thus the results would be entirely biased towards successful projects which makes

them not representative. The website Kicktraq.com is quite useful in that case as it may be used

as a website proxy for all Kickstarter campaigns with disregard of their success. Kicktraq is an

independent platform that provides analysis on current and former Kickstarter campaigns

(Flaherty 2012). To overcome the sample problem, the collection software was programmed to

use Kicktraq first in order to access Kickstarter later, repeating the cycle for each and every

project. Consequently, Kicktraq website is the 1st layer of the data collection design. By doing so,

no loss of information was detected and both successful an unsuccessful projects were

systematically collected.

An ample cross-sectional sample is collected. Only finished projects are selected in order to

assure that the predicted variables, as well as the predictors were accurately represented, and

future variations which might turn out the final project outcome are not possible. Total of 6411

projects have been extracted from Kickstarter.com. However, 6294 projects were used in the

analysis, which stands for 5.9% of all Kickstarter projects ever launched (Kickstarter Statistics Jul

14 2013). The rest 117 records from the ones extracted are either blank records or have key

values missing and thus were not considered. This data failure was caused by software

refreshment on every 24 hours. However, the loss will not have a heavy impact on the

representativeness of the sample, since it stands for barely 1.8% of all the extracted records.

26 | P a g e

As described above, Kickstarter provides a great variety of project types combined into 13

different categories and 50 different sub-categories. Our sample includes all projects from all

categories and sub-categories in the period February 2013 – April 2013. Variable thresholds

have not been set, thus over- or under-represented parameters are not likely. This assures that

the sample should be fully representative and do not suffer from sampling bias.

In the next few sections important navigation elements of the website are described. The

elements that are being used in the analysis are elaborated along with frequently used

Kickstarter terminology.

4.3 Project Landing Page – 2nd layer

The project landing page is the first page of the project the user lands on when navigating

through the website. The project landing page is the first direct link to actual Kickstarter

content. However, due to the data-collection complications stated above it represents the 2nd

layer of the data collection design. Below some key terminology and respectively key elements

from the website are shortly explained.

Backers

Backer is a person who supports the project by funding the same with at least the minimum

amount possible from the funding tiers. This term is adopted in the current thesis.

Pledge and Goal

The funding goal is the money amount announced to be the final target of the project. The

pledged amount is the money amount collected so far. After the period announced for funding,

the project is closed for potential backers. Whether the project is successful or not depends on

the ratio between the pledged amount and the funding goal at the end of the funding period. It

is important to note that the funding goal may be exceeded and respectively the ratio may go

further beyond the value of 1.

27 | P a g e

Facebook integration

In case the user is Facebook integrated within Kickstarter, his or her Facebook network is visible

on the project landing page and user’s activity is usually shared with the network of the user.

For the purpose of the study the number of Facebook friends is only relevant whenever

Facebook integration is available.

External webpages

In the common case this is independent website containing project or creator related

information. When available a reference to the website is visible on the project landing page.

Funding tiers

Funding tiers is defined as the pledge levels that the potential backer can select from when

considering to fund a project. Funding tiers often have “rewards” associated with each one of

them, such as the higher the tier the higher the reward.

Visual reference of the project landing can be found in Appendix - 10.

4.4 Profile page – 3rd layer

Distinct Kickstarter section is dedicated to each user’s profile, including detailed information

about the creator past experience, biography, and general information. This section is

particularly important for the study, since most of the individual-level variables are collected

from here. The Creator’s profile page represents the 3rd layer of the research data collection

design.

Previously created projects

Previously created projects are immediately available at the creator’s profile page. Relevant for

the study are all previously successful and failed projects. Respectively, this information is

collected in a systematic manner for each creator’s profile.

28 | P a g e

Previously backed projects

The website provides full information about each previously backed project. However, the study

is exclusively interested in the number of backed projects, regardless of whether they were

successful or not.

Creator Comments

That section includes all comments made by the user, the respective date and the project that

has been commented on. The study is interested in the number of comments made by the

creator.

Registration date

Finally, at each user’s profile page the respective date of registration is visible. It is scrapped per

user and then converted into continuous variable for the purposes of the analysis.

Visual reference of the user profile page can be found in the Appendix 11.

5 Measures

In order to measure the effect of all factors on the project outcome, data scrapped directly from

Kickstarter is used. Raw data directly collected from the website is complete and comprehensive

measure for some concepts. However, proxies are used for others in order to cover the

complexity of the factors as much as possible. Some concepts are quantified by one measure,

whereas for others multiple measures are used. In the current section the measures used to

construct each factor from the conceptual model are discussed. As noted before, the conceptual

model discriminates between individual-focused factors and project-focused factors. Similar

separation is applied for the measures. Summary of the measures used is presented in the table

below.

29 | P a g e

Table 2 - Summary of the measures

5.1 Project Success

The project success is the main dependent construct in the conceptual model. It is measured as

a ratio between the amount pledged and the funding goal at the end of the funding period

(Success_Ratio). Outcome ratio below the value of 1 indicates non-successful funding, whereas

outcome ratio of 1 or above indicates successful funding. Having continuous measurement of

the success allows for comprehensive view, beyond the mere success of the project. By having

continuous measurement the model avoids giving equal weights to projects that were barely

reaching the funding goal and those that were far exceeded it (being the case of binary

variable). Similarly it avoids equalizing projects that were almost funded with those that were

not funded at all. In this way a precision in measurement is gained which allows for dynamics in

the dependent variable. It is worth mentioning that most of the crowdfunding platforms do not

use “All or nothing model”, but instead adopt “Keep it All” model. Therefore, results based on a

ratio scale for a primary measure of success may have wide range of applications among

different crowdfunding platforms.

However, an important note is that projects with small funding goals receive funding easier,

ceteris paribus. They reach the funding goal and exceed it with relative ease. For this reason the

funding goal is included in the model as a control.

Conceptual Variable: Measurment: Variable Label: Operationalization:

Project success Ratio between the amount pledged and the funding goal Success_Ratio (Pledged_amount)/(Goal)

Number of direct social ties Number of Facebook friends FB_Friends_MCMC Directly observed (continuous)

Reputation Number of previously successful/failed projects Reputation_Index (Previous_Success)-(Previous_Failure)

Community membership Number of days being a member of Kickstarter Membership_Length Directly observed (continuous)

Number of previously backed projects Creator_Backed

Number of previously commented projects Creator_Comments

Presence of external platform/website External_Platform

Presence of financial plan Financial_Plan

Amount of the funding goal Goal

Average amount requested from the funding tires Tiers_Avr

Funding tiers Number of funding tiers related to the project N_Tiers Directly observed (continuous)

Project duration Number of days the project was opened for funding Days_Lasted Directly observed (continuous)

Project age Number of days since the end of the project funding period Days_Since_End Directly observed (continuous)

Project category Project category Games, Music etc. Directly observed (nominal)

Controls/Exploratory

Request size Directly observed (continuous)

Dependent variable

Individual-focused factors

Community involvement Directly observed (continuous)

Project-focused factors

Image credibility Directly observed (nominal)

30 | P a g e

5.2 Direct Social Ties

For measuring the effect of direct social ties the user’s social network information available at

Kickstarter is employed. Once the user is Facebook integrated, and uses Facebook as primary

mean for website access his or her social Facebook network is exposed. We employ number of

Facebook friends of the creator as a proxy measure for direct social ties (FB_Friends_MCMC).

Number of Facebook friends for each creator’s profile is systematically collected when available.

From the sample 69.4 % of the creator’s profiles are Facebook integrated2. Note that direct

social ties assume network outside the pool of people within the online crowdfunding

community. So as the proxy assumes the same. Those are usually connections established prior

to the initiation of the crowdfunding initiative. Nonetheless, the research does not allow for

testing that assumption, and hence the interpretation must be approached with caution.

Despite that, the use of Facebook friends as a proxy for a direct ties is considered

comprehensive, having in mind the social networks boom in the last few years. Facebook

network often represents one’s complete social network.

5.3 Reputation

The role of the reputation is next discussed as a leading driver in the conceptual framework.

Reputation is defined as information about individual’s past performance (Podolny 1994).

Similarly, we quantify reputation as creator’s past performance directly accessed from his

profile. Two measures are used to quantify reputation - number of creator’s successful projects

(Previous_Success) and number of creator’s failed projects (Previous_Failure). Simple reputation

index is computed (Reputation_Index) from the two (see table 2).

2 There is no officially published statistics from Kickstarter on that number, so it is hard to crosscheck the sample with the entire population.

31 | P a g e

5.4 Community Membership

As a measurement of community membership the length of the creator being a member of the

website is used (Membership_Length). This variable represents the number of days the creator

had been a member of Kickstarter at the moment the project finished. The cross-sectional

sample includes collected projects that span over a short period of time (see section 4.2).

Therefore, to avoid bias towards the early projects, the length of the membership is

systematically reduced along the sample with the time between the end date of the project and

the sample collection date.

5.5 Community Involvement

To quantify community involvement the following 2 measures are used: number of projects

previously backed by the creator (Creator_Backed) and number of comments previously made

from the creator (Creator_Comments). The first measure represents the extent to which the

creator is supportive in terms of being financially involved towards the online community. The

second measure represents the extent to which the creator is supportive in terms of being non-

financially involved towards the online community (e.g. sharing ideas, expressing gratitude,

seeking for information, etc.).

5.6 Credibility

Credibility is subjective concept and therefore hard to quantify. It cannot be used in direct

measures, likewise the previous drivers. Instead, proxies are used to capture the effect of that

factor. To quantify image credibility the following proxies are used:

First, the presence of external webpage given as a reference at the user’s profile is used as a

proxy for image credibility (External_Platform). In the common case those are websites

containing project or user related information. It is believed that the existence of external

platform assumes independence of the project and leads to higher credibility of the source.

Binary variable is used to capture this effect. For refining the variable only platforms outside

32 | P a g e

certain domains are considered. Full list of all ignored domains is presented in the appendix (to

be added).

Secondly, the presence of financial plan is used as a second proxy for image credibility

(Financial_Plan). Financial plan implies secure and trustworthy approach for setting the goal

amount. To quantify financial plan binary variable is used. Financial plan is set to be true when

the goal amount is not a round number. In the common case the presence of exact goal amount

assumes having financial plan either supported by an actual one in the project description or

implying one. Note that values such as 999$ or similar have not been considered.

5.7 Request Size

In order to quantify the size of the request the following two measures are used: the funding

goal (Goal) and the average amount requested from the funding tires (Tiers_Avr). For more

details on funding tires see section 4.3. In line with the explanatory nature of the study, the

model is expanded by the number of funding tires attributed to a project (N_Tiers). It is

arbitrary whether this complementary variable captures the effect of request sizes, since it

relates to the variety of the funding anchors, rather than the size of the request itself. For this

reason it is not considered as a measure for request size. However, number of funding tires will

be included in the model both for explanatory and control purposes.

5.8 Dynamics of the Measures

Some measures, related to the individual-level characteristics namely: number of projects

previously backed (Creator_Backed), number of comments previously made

(Creator_Comments), number of previously successful projects (Previous_Success), number of

previously failed projects (Previous_Failure) and Facebook friends (FB_Friends_MCMC) are

dynamic in nature. This means that potential changes may have brought differences between

the values of the same measures by the time the project had ended and the time the sample

had been collected. It is assumed that the potential dynamics are different across different

variables and possibly across individuals and segments. However, these differences cannot be

33 | P a g e

accounted for using a cross-sectional research design. Nonetheless, reducing the time span

between the dependent variable and the above measures limits the potential bias. Therefore,

the sample includes only projects that have ended in the past two months from the date they

were collected. It is believed that two months period is short enough not to lead to some

significant changes that may mislead the results.

6 Results

6.1 Descriptive Statistics

The analyzed dataset contains information about 6289 projects. On average the projects were

active for the window of 32 days (std.dev. = 10.5). The maximum active window was 60 days,

with range 58 days. However, 20% of the projects last for less than 30 days, 50% of the projects

last for 30 days, 30% between 31 and 60 days. On average projects rise an amount of 10 576 $

(std.dev. = 81,915$). The most pledged project raised the impressive 4,188,927$ for 31 days and

outreached his goal 4.7 times. However, not all projects are being successful, since Kickstarter

uses ‘All or nothing model’. The average success rate is 1.12 (std.dev. = 4.19), which is slightly

over the critical ratio of 1, below which the project fails. Roughly 51% of the initiated were

successful and reached their funding goal. 63$ is the average pledge per person (std.dev. = 86$)

and 2600 $ is the maximum amount. On average projects have 153 backers (std.dev. = 1193).

The average goal set is 28,380$, having 4 updates, 38 comments and 9 funding tiers. 67% of the

projects have their own website and 3% have a financial plan. On average initiators have 0.17

previously successful projects, 0.16 previously failed projects, being members from 248 days,

having 743 Facebook friends. Table with descriptive statistics can be found in Appendix 1.

The most initiated projects are in categories Film & Videos and Music, following by Publishing

and Art. A histogram with all project categories and subcategories and their share in the sample

can be found in Appendix 2. The category with the highest average pledge is Technology (100$)

and the lowest is Comics (44$). The most successful category is Dance with 81% of the initiated

projects being funded and Fashion is the most unsuccessful with 35%. To a great extent simple

34 | P a g e

statistics from the sample match with statistics officially published at the Kickstarter website

(Kickstarter Statistics Jul 14 2013). That crosscheck makes the sample robust and representative,

potentially having limited sampling bias.

6.2 Correlation Coefficients

Correlation coefficient matrix is used as an initial sanity check for the underlying conceptual

assumptions. Most of the Pearson correlation coefficients show significant relationship between

variables. However, significance between most of the variables is deemed to be normal,

considering the ample sample size.

The latent variables constructed in the conceptual model are confirmed by the correlation

coefficients. Strong correlations between variables assigned to the same latent factor are: the

size of the funding goal (Goal) and tiers average (Tiers_Avr) (r=0.585, p<0.01); number of

creator backed projects (Creator_Backed) and number of creator comments

(Creator_Comments) (r=0.510, p<0.01).

Number of days being a member (Membership_Length) and number of creator backed projects

(Creator_Backed) are quite strongly correlated (r=0.452, p<0.01). That is expected, since the

longer one is a member of the crowdfunding community the higher the chance to support.

However, being simply a member does not always assume being an active member. This may

explain the moderately strong correlation between the two variables. Nevertheless, it has been

discriminated between membership and involvement and account for interaction between the

two.

The strong correlation between number of comments on the project (N_Comments) and

number of creator comments (Creator_Comments) (r=0.796, p<0.01) reveals important relation.

The strong correlation index assumes most of the comments on the project page are indeed

initiated by the creator. And respectively the majority of the comments related to the creator’s

profile are comments on his own project. That may include, but is not limited to answering

users’ questions, requests, updates on the current project status (beyond the official update

section), triggering discussions, etc. This finding is not in line with the concept of community

35 | P a g e

involvement developed in the conceptual model. It seems that by commenting a creator gets

involved primarily in his own project, and less in other community projects. Therefore, it is

disclosed that the creator’s comments are not a proper measure for community involvement,

thus is being dropped as such.

A strong correlation was found between number of project updates (N_Updates) and number of

creator comments (Creator_Comments) (r=0.564, p<0.01). It seems that in contribution to

updating on the project wall, the creator makes sure that the update is communicated properly

to all, by triggering discussions in the comments section. Having in mind that project updates

are a strong confounding variable (see correlation matrix in Appendix - 3), communicating them

in various ways looks like best practice in managing a crowdfunding project.

A full table of all correlations coefficients can be found in Appendix - 3.

6.3 Regression Models

6.3.1 Validity and Assumptions

Before running the regression model the data was checked for the underlying linearity and

normality assumptions, refined from outliers and checked for missing values and

multicollinearity issues. The guidelines of Hair et al. (2010) are followed.

6.3.1.1 Normality Assumptions

Normality assumptions are test which empirically examine the shape of the distribution and Q-Q

plots for each variable as well as the values for kurtosis and skewness. Normality tests were

following the guidelines of Hair et al. (2010). From the skewness values it becomes clear that all

variables are heavily positively skewed. After visual inspection of the distribution shape the

same was confirmed. The severe non-normality was partly solved using natural logarithm

transformation, suggested by Hair et al. (2010). To overcome logarithm of zero and to avoid

negative values a constant of 1 is added. In this way the model adapts to the widely accepted

36 | P a g e

ln(X+1) transformation on both sides of the equation. Subsequent to the transformation the

regression coefficients have to be interpreted as elasticities. After testing the transformed

variables there was a significant improvement in the distributions. However, the Kolmogorov-

Smirnov tests failed to reject the null hypothesis even after the transformation. This is expected

though, since for large sample sizes strict normality tests such as Kolmogorov-Smirnov often fail.

Overall, after the transformation most of the variables show no significant deviations from

normality. The relative non-normality for some variables is neglected, considering the large

sample size. Tables for Skewness and Kurtosis values before and after the transformation, as

well as distribution histograms and Q-Q normality plots after the transformation can be found in

Appendix 4-5.

6.3.1.2 Linearity Assumptions

Linearity assumptions have been tested examining the residuals normality, as suggested by Hair

et al. (2010). Before the logarithm transformation regression residuals have significant

deviations from normality. This is shown on the distribution histograms and normal P-P residual

plots. After the transformation, normality is decidedly improved. Any deviations from normality

after the transformation are neglectable and linearity assumptions are confirmed. Residual

histograms and P-P plots before and after the transformation can be found in Appendix 6.

6.3.1.3 Outliers Detection

Due to the log transformation extreme values were significantly reduced. However, some

observations might still cause problems when analyzing the data and we want to make sure

they are not considered in the analysis. The discussion is based on two broadly applied methods

for detecting outliers - outlier labeling rule proposed by Tukey (1977), including the

forthcoming modifications of the applied method by Hoaglin (1987) and the deviations from the

mean examined by the z-scores of the standardized values. For choosing the right method we

investigate for outliers on the dependent variable –Success ratio. Given the fact the

distributions are positively skewed only the upper bound outliers are a point of interest. The

37 | P a g e

outlier labeling rule is used: Q3+g*(Q3-Q1), where Q1 and Q3 are respectively the 25th and the

75th percentile, and ‘g’ is coefficient equals to 2.2 (as discussed by Hoaglin 1987). After the

results the dependent variable of Success Ratio has 68 extreme values. When examining the z-

scores of the standardized values (for z<3.29) success ratio has 85 extremes. It is clear that both

tests do not show any major differences. Dealing with outliers is often based on the author’s

subjective judgment. Therefore, it was decided that Outlier labeling rule with g =2.2 will be used

for all variables in the dataset and the data will not be refined extensively. After the data

cleaning 171 cases were deleted, accounting for 2.72% of the original sample size.

6.3.1.4 Missing Values

The only variable that has missing values is Number of Facebook friends. About 69% of the

observations have valid values for that variable. This is due to the fact that not all Kickstarter

users have Facebook integration on their account. Including the variable with the missing values

in the regression model would substantially reduce the sample size. The research aims at

capturing the effect on direct ties over the project outcome and FB friends are just a measure to

achieve that. The missing values do not assume that direct ties are absent for these

observations, they are just not been observed. For this reason the missing values were

simulated by Monte Carlo Markov chain (MCMC) method for data imputation. Many studies

show no significant difference between MCMC and other imputation methods such as

expectation-maximization (EM) algorithm (see Lin (2010)). However, predictive models based

on imputed data are generally considered to be more precise than those based on non-imputed

data (e.g. replace with the mean). Using the MCMC method, the missing data is considered to

be effectively replaced, and thus the sample size is used at its full capacity.

6.3.1.5 Multicollinearity

The regression model was inspected for multicollinearity examining variance inflation factors

(VIF). The guidelines of Hair et al. (2010) were used and all the variables in the models have VIF

values below the critical threshold of 10. The variables were also checked for direct correlations

(see section 6.2). As VIF levels are below the critical values and no strong correlations were

38 | P a g e

found between the covariates, it was concluded that the regression models were not affected

by multicollinearity problems.

6.3.2 Multiple Models

The current section summarizes the results from the two regression models. In the forwarding

subsections each model is examined individually. Summary table with the results from both

models is presented below.

Table 3 - Multiple Models

Model 0 (controls) Model 1 (controls+IV) Model 2 (controls+IV+Reputation)

Standardized Beta Sig. t



(Constant) .000 13.132 .000 13.328 .029 2.186

LN_Days_Since_End -.025 .038 -2.079 -.017 .103 -1.632 -.015 .150 -1.439

LN_Days_Lasted -.149 .000 -12.213 -.060 .000 -5.391 -.056 .000 -5.092

LN_N_Tiers .248 .000 19.967 .257 .000 18.982 .254 .000 18.851

Games .019 .150 1.438 .021 .086 1.716 .022 .078 1.764

Music .085 .000 5.628 .025 .070 1.814 .020 .138 1.484

Publishing -.027 .063 -1.861 -.059 .000 -4.602 -.060 .000 -4.668

Design .032 .016 2.406 .040 .001 3.284 .041 .001 3.358

Fashion -.036 .005 -2.791 -.050 .000 -4.296 -.047 .000 -4.039

Art .033 .019 2.354 -.022 .076 -1.776 -.023 .066 -1.838

Technology .011 .382 .875 .055 .000 4.778 .054 .000 4.756

Theater .045 .000 3.494 .026 .020 2.329 .023 .042 2.037

Comics .103 .000 8.072 .036 .002 3.124 .038 .001 3.350

Photography -.009 .460 -.739 -.042 .000 -3.783 -.040 .000 -3.616

Food -.001 .950 -.062 .020 .092 1.684 .019 .093 1.682

Dance .057 .000 4.634 .028 .009 2.605 .027 .012 2.509

LN_FB_Friends_MCMC .113 .000 10.193 .107 .000 9.660

LN_Membership_Length

.035 .004 2.863 .035 .003 2.931

LN_Creator_Backed

.248 .000 19.357 .238 .000 18.615

LN_Goal

-.409 .000 -28.151 -.411 .000 -28.402

LN_Tiers_Avr

.018 .253 1.144 .020 .190 1.310

External_Platform

.087 .000 7.835 .082 .000 7.436

Financial_Plan

-.015 .148 -1.446 -.014 .178 -1.348

LN_Reputation_Index

.093 .000 8.621

F 49.006 123.428 122.713

df 6117 6117 6177

R Square .108 .308 .317

Adjusted R Square .105 .306 .314

39 | P a g e

6.3.2.1 Model 1 Results

The first model includes all variables in the conceptual model and the variables controlled for

the validity of the results. In the model 14 controls are added as follows: 12 project categories,

project age (Days_Since_End) and number of days the project was opened for funding

(Days_Lasted). Note that the number of project updates, the number of project comments and

creator’s comments (as being highly correlated to project comments, see section 6.2) are not

included as controls to avoid overmatching and statistical bias (LaLonde 1986). The

hypothesized mediator (Reputation_Index) was not included into that first model. The model is

expanded later with the mediator and the effects are discussed accordingly.

The regression equation of the linearized first model is as follows:

ln(Success_Ratio)=0+1ln(FB_Friends)+2ln(Membership_Length)+3ln(Creator_Backed)+4ln(Go

al)+5(Website)+6(Financial_Plan)+7ln(Tires_Avr)+8ln(N_Tiers)+9ln(Days_Since_End)+10ln(Day

s_Lasted)+11(Games)+12(Music)+13(Publishing)+14(Design)+15(Fashion)+16(Art)+17(Tech

nology)+18(Theater)+19(Comics)+20(Photography)+21(Food)+22(Dance)+e

The predictors in Model 1 significantly differ from zero (F=123, p<0.01), therefore the model has

meaningful statistical interpretation. The adjusted R square indicates that 31% of the variation

in the Success Ratio is explained by the variation in the covariates included into the base model.

This is satisfactory result considering the measurement limitations and the technological

constraints. Taken together the research aims to explain key individual and project-related

factors and their effect on the project success and not to construct a full predictive model. It is

accounted for the unexplained variation to endogenous factors that cannot be measured or are

hard to be quantified (e.g. video appeal, cause dedication, project eagerness, fashion etc.).

Summarizing the base model six out of eight key variables have significant effect on the project

success. The most influential variable is the funding goal (Beta=-0.409, p<0.01). Financial plan

and average tiers amount are not significant predictors. All covariates and the respective

coefficients are listed in Table 3. Further discussion on the results is presented in section 7.

40 | P a g e

6.3.2.2 Model 2 Results

In the second model next to the direct effect of social ties (FB_Friends_MCMC) on the project

success the indirect effect through the community reputation (Reputation_Index) is included.

Quantifying direct social tiers and community reputation is explained in chapters 5.2 and 5.3.

Multiple regressions are used to test the mediation model following the guidelines of Hair et al.

(2010) and Baron (1986). However, due to statistical limitations one should know that successful

mediation may be overlooked using this method (see Baron (1986)). Nonetheless, this is

classical approach and the same is used for the current thesis. First, the model was tested for

established significant correlations between the constructs. All constructs are significantly

correlated as shown in the correlation matrix (see Appendix 3). Next, the model was tested for

significant effect of direct social ties (FB_Friends_MCMC) as an input covariate on community

reputation index (Reputation_Index) as a predicted variable. In correspondence with the base

model (Model 1), it is controlled for all other variables. The number of Facebook friends are

shown to be significant predictor for community reputation (Beta=0.068, p<0.01). Finally the

initial model (Model 1) was estimated with the community reputation index included to build

the secondary model (Model 2). The overall model explanatory power has increased (Adjusted R

Square = 0.314) and the model significantly differs from zero (F=123, p<0.01).

The effect of Facebook friends on the project success rate is weakened for the second model

(Beta=0.107, p<0.01) when controlled for the community reputation, as compared to the first

model (Beta=0.113, p<0.01). However, in both models Facebook friends remain significant

predictor on the project success rate. This assumes existence of partial mediation effect of

community reputation (Reputation_Index) on the effect of direct social ties (FB_Friends_MCMC)

on the project outcome. Further discussions are followed in section 7.2.

41 | P a g e

7 Discussion on the Results

The discussion is based on the results from Model 1 and Model 2 in table 3. Additionally

interaction terms are discussed based on complementary models embedded to the appendix.

Direct effects from Model 1 are discussed in the sections 7.1, 7.3, 7.4 and 7.5. Direct effects and

mediation effects of community reputation from Model 2 are discussed in section 7.2. Project

categories and interaction terms are discussed in section 7.6.

7.1 Direct Social Ties

From the base model (Model 1) results it is clear that the number direct social ties

(FB_Friends_MCMC) have fairly high direct positive impact on the project success ratio

(Beta=0.113, p<0.01). The observation is in line with hypothesis H1, hence the same is accepted.

This is possibly because information asymmetry is the main obstacle for entrepreneur to obtain

external funding and network tiers are important instrument for overcoming information

asymmetry (Shane 2002). In that respect it looks particularly important for a crowdfunding

entrepreneur to utilize the power of social network connections, in order to overcome the

limited knowledge. Sharing among the direct social network may also create a buzz effect

contributed by the 2nd-level, 3rd-level and so forth networks. For that being possible the activity of

the network members is important for effective blowout of the message. Therefore,

communication beyond simple online social share is necessary for funding activity to be

triggered.

7.2 Reputation

In the secondary model (Model 2) the addition of the reputation index (Reputation_Index)

indicates fairly strong positive relation with the project outcome (Beta=0.093, p<0.05). A

conclusion may be drawn from this that higher reputation index is associated with higher

success in the crowdfunding initiative. Hence hypothesis 2a is accepted. Upon decomposition of

the reputation index it seems that previous failure, as a measure for negative reputation

42 | P a g e

weights more (Beta=-0.110, p<0.01) than previous success (Beta=0.098, p<0.01). This means

that funding interest is withdrawn easily, when the creator has negative reputation, as

compared to the strengthen effect for a creator with positive reputation. That finding is in line

with previous studies that the effect of negative as opposed to positive information is stronger

(Fiske 1980). The outcome of this result asserts certain implications on the extend towards

which the project creator should give publicity to previous projects. As an example previous

successful projects should be used as a reference point for attracting potential investors when

promoting a crowdfunding project. In this way the credit is given to the positive creator’s

reputation which should signal the entrepreneur’s strong skills and expertise. On the contrary,

publicity to previous failed projects should be limited or given with caution.

The effect of direct social ties (FB_Friends_MCMC) on the project success rate is weakened for

the secondary model (Model 2) (Beta=0.107, p<0.01) when controlled for the community

reputation, as compared to the base model (Model 1) (Beta=0.113, p<0.01). In both models

(FB_Friends_MCMC) the statistically significant predictor remains and the effect of reputation

does not fully displace that relationship. The relative drop in the coefficient for the second

model is rather small - 5.3%, hence it is hard to claim that strong mediation exists. Nonetheless,

the results support the existence of partial mediation of community reputation on the effect of

direct social tiers on the project outcome (see section 6.3.2.2). Hence, hypothesis 2b is accepted

with the remark that we observe partial mediation effect. It seems that direct social ties are

steady predictor of the project success and reputation does not overshadow their effect. This

assumes that reputation acts mainly as a signaling variable for potential investors and has less

influence on the information transfer.

7.3 Community Membership and Involvement

7.3.1 Community membership

The length of the community membership (Membership_Length) is fairly weak but significant

predictor of the crowdfunding success (Beta=0.035 p<0.01). Hence, hypothesis H3 is accepted.

The results enforce the idea that group-oriented behavior exists within the crowdfunding

43 | P a g e

community. Indeed, in-group favoritism occurs between members of the community and senior

members seem to be favorably treated. It is likely that senior members have adopted social

norms inherent to the community, and as such are being perceived as more valuable members.

That asserts certain implications for potential crowdfunding entrepreneurs. They should

consider being members of the community enough before initiating a project. Otherwise they

would risk being perceived as not consistent members which might potentially reduce the

chance to be funded by seniors in the community.

Being simply a senior member does not always assume being an active member as well.

However, when tested for interaction terms the effect of community membership

(Membership_Length) is not significantly moderated by the community involvement

(Creator_Backed) (Beta=-0.100, p=0.114). To avoid multicollinearity between the interaction

product term and the independent variables, the same were mean-centered. Despite this the

results are consistent and no significant interaction effect was observed in either cases. The

regression table can be found in Appendix 7. The results demonstrate that the two main effects

act independently and do not interact each other.

7.3.2 Community involvement

Community involvement is a factor aimed at being quantified by two observed variables –

previously backed projects (Creator_Backed) and previously commented projects

(Creator_Comments). However, as explained in section 6.2, it seems that in the crowdfunding

community the creator mostly comments on his own projects, and rarely on others’ projects,

which is not in line with the concept of community involvement. As a result the measure is not

considered in the analysis and therefore the non-financial involvement of the member toward

other community members is not explored.

However, the financial involvement in the community quantified by previously backed projects

(Creator_Backed) seems to be the third most influential factor (Beta=0.248, p<0.01). Once

again, the results support the idea of in-group favoritism. Members of the crowdfunding

community are more favorably inclined to financially support other members if the last have

44 | P a g e

been financially active in the community before. Hence, hypothesis H4 is accepted. This finding

asserts some practical implications as well. Crowdfunding members need to consider supporting

one or more projects prior to an initiation of their own. Doing so will trigger favorable behavior

and empathy from the other members and potentially will lead to a better outcome for their

own project.

7.4 Credibility

The presence of a financial plan (Financial_Plan) turns out to be an insignificant predictor for the

crowdfunding success (Beta=0.015, p>0.05). As such it is not considered as a measure for

credibility. However, this is a surprising result, since financial planning has deemed to be an

important determinant for forming a credible image of the project. One potential explanation

might be that barely 3% of the projects in the sample size have financial plan, thus the

observations are not sufficient for the coefficient to reach significance. Another more

elaborative explanation is the way the financial plan is communicated. Having a non-round

number might indicate a detailed plan on the funds needed to implement the crowdfunding

idea. It is implied on the project page. However, if not stated explicitly, the potential donor

might never consider the existence of such. An important managerial conclusion should be that

whenever having a plan, financial or otherwise it should be communicated properly.

On the other hand the presence of an external platform (External_Platform) used as a project

reference seems to be significant predictor of crowdfunding success (Beta=0.087, p<0.05).

Hence, hypothesis H5 is accepted. The results might indicate that potential investors often seek

information outside the crowdfunding platform and refer to external webpages. In this respect

having a website, or a blog turns out to be important for potential funders to verify the

credibility of the source and to maintain trustworthy image. From a managerial perspective it is

worth considering higher exposure to any external sources of information related to the

creator, the project or both.

45 | P a g e

7.5 Request Size

Average amount requested from the funding tiers (Tiers_Avr) is not significant predictor for the

crowdfunding success (Beta=0.018, p>0.05). However, the number of funding tiers (N_Tiers)

have strong significant effect on the project success rate (Beta=0.257, p<0.01). It seems that it is

not the amount requested from the funding tiers, but rather the variety of funding levels that

matters. In fact when the number of funding tiers is not controlled for, the average request is

statistically significant positive predictor (Beta=0.145, p<0.01). Thus the average amount

requested is fully mediated by the number of funding tiers. The results assume that one may

rarely think that the project funding tiers are set high, as long as different levels of support exist.

It is likely though, potential funders to be tentative in their decision to support once they do not

have the proper funding tier. Crowdfunding entrepreneurs should consider variety of support

levels available for their projects, from token support to substantial financial backup.

The funding goal (Goal) is the most influential predictor in the model (Beta=-0.409, p<0.01). It

shows that projects with lower funding goals tend to be more successful on average than

projects with higher funding goals. Since projects with lower goals are naturally more likely to

reach and outreach the funding goal the results were crosschecked by the implementation of a

binary variable for the project success. By using logistic regression model consistent results

were observed for the effect of the funding goal on the binary dependent variable (B=-0.692,

p<0.01). The logistic regression table can be found in Appendix 8. Since the funding goal is a

primary measure for the request size hypothesis H6 is accepted. This leads to the assumption

that higher request size may be perceived as unreasonable, thus investment interest can be

withdrawn. In setting the funding goal one must consider significant, yet reasonable target. If

possible best practice is to set smaller projects with reasonable goals in a roll rather than set

high optimistic target at once. Also widely used are the so called “stretch goals” (Kickstarter

Webpage 2013). Those are targets beyond the initial goal if the same is reached. Important

managerial implications can be drawn from the above results and potential crowdfunding

entrepreneur should always consider these practices when defining the funding goal of their

project.

46 | P a g e

7.6 Project Category

Analysis shows that technology projects are on average more successful than films & video

projects (Beta=0.055, p<0.01). The same applies also for design, comics, dance and theater. On

the contrary fashion and photography projects are significantly less successful than films &

video. Games, music, art and food do not differ significantly from video & films category. These

results reveal important information for potential crowdfunding entrepreneurs in photography

and fashion projects. They should consider the relative difficulty in obtaining funds from the

crowd and try to optimize their projects.

Possible moderation effects are explored based on some strong correlations between project

categories and independent variables (see Appendix 3). To avoid multicollinearity the variables

were mean centered. The full model is presented in Appendix 9. Games category is significant

moderator on the number of funding tiers (N_Tiers). This results in game projects with variety of

funding levels being significantly more successful (Beta=0.091, p<0.05) on average. These results

assume that the variety of funding levels is even more important when having a gaming project.

The same applies for the community involvement (Creator_Backed) (Beta=0.041, p<0.05).

Supporters that fund gaming projects seem to be especially sensitive to the creator’s past

financial involvement in the community. Lastly games are significant moderator on the project

goal (Beta=-0.207, p<0.01). This shows that many game supporters do not tolerate high goals,

so potential entrepreneur in that category should be especially interested in the best practices

presented in the previous section.

8 Conclusion

8.1 Research Relevance

The development of online technologies and more particularly the extension of Web 2.0 have

made possible typical offline activities to be brought online. This peculiar transformation has

47 | P a g e

made fundraising possible in an online setting. Nowadays, entrepreneurs instead of approaching

banks, venture capitalists or business angels, set online “open calls” in support of their projects.

This phenomenon has been referred to as “crowdfunding” or financial support from the public

or the ‘crowd’ via internet. The new alternative becomes particularly popular among

independent music projects in the beginning of 2006. However, in a few years entrepreneurs

see the opening possibilities and now crowdfunding is used for various types of projects from

film & video to technology and design. In article from Caldbeck (2012) it has been claimed that

the crowdfunding industry reached $2.8 billion in equity. Nowadays hundreds of crowdfunding

platforms exist and their number is exponentially growing. The rapid development stresses the

need to better explore the crowdfunding phenomenon and to understand the determinants of

success. The current research provides new insights on the crowdfunding, based on empirical

findings from the biggest crowdfunding platform in world – Kickstarter.

8.2 Summary of the Findings

The current study adopted literature from marketing, consumer behavioral, managerial and

social behavior sciences. A conceptual model is created based on two general classifications:

individual-focused factors, which relate to characteristics of the entrepreneur (social ties,

reputation, and membership and community involvement) and project-focused factors (image

credibility and request size). Additionally, the role of the project categories and existing

moderations are explored.

The empirical analysis is based on 6289 projects from the biggest crowdfunding platform

Kickstarter.com. The data is indirectly extracted from Kickstarter through the use of 3rd party

platform Kicktraq.com. The data collection design consists of 3 layers through which metrics

about each project have been scrapped.

The results reveal important aspects of the crowdfunding outcome, with possibly high

managerial applications. The most influential predictor is the project goal, which shows that

supporters are highly sensitive to the request size. In addition, more funding levels are better

48 | P a g e

accepted on average than less funding levels. That is especially important when initiating a

gaming project, as that category shows to be significant moderator on the funding tiers.

Direct social ties are also significant predictor on the project outcome. The results show that the

direct connection network of the entrepreneur is useful tool for overcoming the existing

information asymmetry. Moreover, it seems the utilization of the online buzz through social

networks is important for crowdfunding entrepreneurship.

The role of the reputation has positive direct effect on the project success. Entrepreneurs with

previously good community reputation tend to be more successful. Reputation turns out not to

be a strong mediator of the direct network ties, contrary to what was hypothesized. It can be

concluded that next to the previously successful record, the role of the social ties is indeed

important.

Community membership and involvement are also important determinants of success. The

longer the entrepreneur is a member of the crowdfunding platform, the better the chances for

a favorable outcome. Similarly, supporters are more favorable towards entrepreneurs that have

been financially involved in previous crowdfunding projects. All this supports the idea of in-

group favoritism.

The presence of financial plan is not significant predictor of the project outcome and is hard to

be used as a measurement for image credibility. Reasonable explanation might be that financial

plans are often not communicated properly, so potential supporters do not consider them. The

presence of external webpages, however, gives credibility to the source and affects the decision

of the potential investor.

8.3 Limitations and Future Research

The conceptual model used in that research is empirically tested entirely by quantitative data.

By doing so, some important insights which are otherwise influential drivers for the

crowdfunding success might be concealed. Therefore, the use of qualitative research is

49 | P a g e

recommended to crosscheck the results and the underlying assumptions in order to uncover

potential dependencies within the framework.

The statistical model has relatively low exploratory power (31%). This clearly indicates that

factors more influential than the ones explored are driving forces behind every successful

crowdfunding campaign. Experience shows that many supporters are influenced by the video

and picture materials related to the project. Others might be entirely dedicated to the good

cause of the project. Third are fascinated by the eagerness of the creator, thus feel the need to

support his creativity. Fashion trends might also be important predictors. All these factors

possibly interact with each other as well. Unobserved in the current study, they leave room for

further research in this direction.

Kickstarter is the biggest crowdfunding platform. Since the results are based on a broad sample

from the platform they are considered generalizable for other platforms as well. However,

Kickstarter does not allow charity projects and personal causes. The underlying motives for such

projects would be entirely different, so as the supporting behavior. In this respect the

applications from the current study should be interpreted with caution when it comes to charity

projects. However, similar research framework may be applied to platforms supporting charity

projects (e.g. startsomegood.com, causes.com) in order to reveal differences in the funding

behavior.

The current study uses cross-sectional sample from the crowdfunding platform which provides a

snapshot of the projects at the end of the funding period. It would be valuable from a

managerial perspective to follow timeline dynamics that uncover insights in the project funding

lifecycle.

8.4 Research Contributions

As noted in the beginning the current paper adds up to the research from Evers (2012) and

expands his findings by explaining the individual-based factors related to the project

entrepreneur, namely direct social ties, community reputation, membership and involvement.

50 | P a g e

To the best of our knowledge no research so far has examined these aspects of crowdfunding,

so it is believed that the current paper contributes to the existing academic lit. Project-related

factors such as the project goal and the number of funding tiers are covered by Evers (2012).

The current paper confirms the findings from Evers (2012) with regard to the project funding

goal. However, the current research found discrepancies in the effect of the number of funding

tiers on the project success. Significant positive effect was discovered, whereas Evers (2012)

reveals the opposite. These differences are most probably attributed to the different research

platforms (as Evers (2012) adopts IndieGoGo) used by the two papers and respectively the

differences in the funding behavior. Such dissimilarities should stress the need for diversification

in the crowdfunding research, as apparently variety exists and the same should be examined.

8.5 Managerial Implications

Many entrepreneurs find viable alternative to traditional funding means in the face of

crowdfunding. Furthermore, the exponential growth of existing crowdfunding platforms provide

even bigger and diverse opportunities for those seeking to kick start their ideas. The relative

simplicity of crowdfunding makes this new funding mean even more appealing for

entrepreneurs and especially for projects smaller in scale. Nevertheless, one should realize that

running successful crowdfunding project is not an easy task. It needs proper planning, devotion,

entrepreneurial mindset and creative approach.

The current thesis provides valuable managerial insights with regards to the research points

examined. Everyone who considers using crowdfunding as a funding alternative should utilize

the power of the social network connections. Depending on the crowdfunding platform most of

them provide social network integration. This functionality might be especially useful for

synchronizing and sharing all the activities with online social networks and helps to create a

communication stream for potential investors.

Previous crowdfunding experience might be a strong indication for possessed skills and

expertise. However, negative experience would more likely withdraw investment interest to a

bigger extent than the uplift from the positive experience. In that respect the entrepreneur

51 | P a g e

should raise publicity only to previously successful projects, and should try to limit the public

exposure of failed project initiatives. It is worth mentioning that such measures are already

taken by Kickstarter, as they restrict the exposure of failed projects and in turn increase the

exposure of successful ones.

Before initiating a crowdfunding project one should know that the crowdfunding platforms are

often seen as online communities, and as such members often behave favorably towards each

other and unfavorably towards the outsiders. In that respect adopting social norms and

community integration are important steps one should take before starting a project.

Involvement in community projects and supporting other’s initiatives are worth considering

from a prospective crowdfunding entrepreneur. Creating the positive image of a healthy

member also requires time that should be planned in advance.

For conveying a credible image, the entrepreneur has to provide external sources of information

about his project or ideas, such as other websites, blogs or external groups. However, this has to

be approached with caution since such sources sometimes can create negative image of the

entrepreneur. Therefore, external feeds should be carefully controlled and utilized whenever

possible.

When setting the funding goal the entrepreneurs should know the importance of the task, as

this factor is essential determinant for the project success. One must consider significant, yet

reasonable target. This target is highly individual for every project and is hard to create a single

common rule that applies for all. Nevertheless, this research points out the general trend that

projects with higher goals tend to be on average less successful. There are already well

established best practices in the field such as the use of “stretch goals” or crating few smaller

projects in a roll. It is important, however, that the project entrepreneur becomes acquainted

with these and plans his strategy beforehand. Furthermore, when considering Kickstarter as a

crowdfunding platform providing higher decision control by the investor by increasing the

funding levels is a better strategy than guidance and limited funding levels.

52 | P a g e

9 References:

Agrawal, A., C. Catalini and A. Goldfarb (2011). "Online Relationships, Distance, and the Internet: The Geography of Crowdfunding." The National Bureau of Economic Research NBER Working Paper No. 16820.

Bagozzi, Richard and Dholakia, Utpal (2002). "Intentional social action in virtual communities." Journal of Interactive Marketing 16(2): 2-21.

Bagozzi, Richard P. and Moore, David J. (1994). "Public Service Advertisements: Emotions and Empathy Guide Prosocial Behavior." Journal of Marketing 58(1): 56-70.

Baron, David A. Kenny and Reuben M. (1986). "The Moderator-Mediator Variable Distinction in Social Psychological

Research: Conceptual, Strategic, and Statistical Considerations." Journal of Personality and Social Psychology 51(6): 1173-1182.

Bayus, Venkat Kuppuswamy and Barry (2013). "Crowdfunding Creative Ideas: The Dynamics of Project Backers in Kickstarter." March 17, 2013 (working paper).

Belleflamme, Paul, Thomas Lambert and Armin Schwienbacher (2010). "Crowdfunding - An Industrial Organization Perspective." Digital Business Models: Understanding Strategies: pp. 25–26.

Belleflamme, Paul, Thomas Lambert and Armin Schwienbacher (2011). "Crowdfunding -Tapping the Right Crowd." Center for Operations Research and Econometrics, Louvain-la-Neuve, Belgium.

Bendapudi, Neeli, Surendra N. Singh and Venkat Bendapudi (1996). "Enhancing Helping Behaviour - An integrative Framework for Promotion Planning." Journal of Marketing 60(3 ): 33-49.

Bhide, Amar V. (1992). "Bootstrap finance: the art of start-ups." Harv Bus Rev 70(6): 109-117.

Birley, Sue (1987). "New ventures and employment growth." Journal of Business Venturing 2(2): 155–165.

Blanca, Sebastian (2012). Top Crowdfunding Sites for Designers, Jun 17, 2012. crowdsourcing.org.

53 | P a g e

Brockner, Joel, Beth Guzzi, Julie Kane, Ellen Levine and Kate Shaplen (1984). "Organizational Fundraising: Further Evidence on the Effect of Legitimizing Small Donations." Journal of Consumer Research 11(1): 611-614.

Burnkrant, Robert and Page, Thomas (1982). "An Examination of the Convergent, Discriminant, and Predictive Validity of Fishbein's Behavioral Intention Model." Journal of Marketing Research 19(4): 550-550.

Caldbeck, Ryan (2012). "Crowdfunding Predictions for 2013." from http://www.forbes.com/sites/ryancaldbeck/2012/12/11/crowdfunding-predictions-for-2013/.

Chong, Wang (2012). Network Centrality and Contributions to Online Public Good--The Case of Chinese Wikipedia.

Dasgupta, Nilanjana (2004). "Implicit Ingroup Favoritism, Outgroup Favoritism, and Their Behavioral Manifestations." Social Justice Research 17(2): 143-169.

Dell'Ariccia, Giovanni and Enrica Detragiache and Raghuram Rajan (2008). "The real effect of banking crises." Journal of Financial Intermediation 17(1): 89–112.

Dholakia, Utpal M., Richard P. Bagozzi and Lisa Klein Pearo (2004). "A social influence model of consumer participation in network- and small-group-based virtual communities." International Journal of Research in Marketing 21(3): 241-263.

Ebben, Jay, and Alec Johnson (2006). "Bootstrapping in Small Firms: An Empirical Analysis of Change over Time." Journal of Business Venturing 21, (6): 851-865.

Ellemers, Naomi and Kortekaas, Paulien and Ouwerkerk, Jaap (1999). "Self-categorisation, commitment to the group and group self-esteem as related but distinct aspects of social identity." European Journal of Social Psychology 29(2-3): 371-389.

Evers, Mart (2012). Main drivers of crowdfunding success: a conceptual framework and empirical analysis, Rotterdam School of Management, Erasmus University.

Fiske, Susan T. (1980). "Attention and weight in person perception: The impact of negative and extreme behavior." Journal of Personality and Social Psychology 38(6): 889-906.

Flaherty, Joseph (2012). "Kicktraq Is Like Google Analytics for Kickstarter." from http://www.wired.com/design/2012/08/kicktraq-google-analytics-for-kickstarter/.

http://www.forbes.com/sites/ryancaldbeck/2012/12/11/crowdfunding-predictions-for-2013/

http://www.wired.com/design/2012/08/kicktraq-google-analytics-for-kickstarter/

54 | P a g e

Hair, Joseph, William C. Black, Barry J. Babin and Rolph E. Anderson (2010). "Multivariate Data Analysis." Pearson Prentice Hall 7th Edition.

Hoaglin, D. C., and Iglewicz, B. (1987). "Fine Tuning Some Resistant Rules for Outlier Labeling." Journal of American Statistical Association(82): 1147-1149.

Howe, J. (2006). The rise of crowdsourcing Wired Magazine. Issue 14.06.

Hughes, Andy Cosh & Douglas Cumming & Alan (2009). "Outside Entrepreneurial Capital." Economic Journal, Royal Economic Society vol. 119(540): pages 1494-1533.

Johnsen, Becky (2010). "Information Asymmetry." Econ Focus 4(14).

Kappel, T. (2009). "Ex Ante Crowdfunding and the Recording Industry - A model for the US." Loyola of Los Angeles Entertainment Law Review 29(3).

Kelman, Herbert C. (1961). "Processes of Opinion Change." The Public Opinion Quarterly 25(1): 57-78.

Kickstarter Statistics (Jul 14 2013). kickstarter.com/help/stats.

Kickstarter Webpage (2013). "http://www.kickstarter.com/help/faq/kickstarter%20basics." from http://www.kickstarter.com/help/faq/kickstarter%20basics.

Kleemann, Frank, Kerstin Rieder and G. Günter Voß (2008). "Un(der)paid Innovators: The Commercial Utilization of Consumer Work through Crowdsourcing." Science, Technology & Innovation Studies 4.

Kuile, Franciscus Alexander ter (2011). The State of Crowdfunding: A Review of Business Models and Platforms, Universiteit van Amsterdam.

LaLonde, Robert J. (1986). "Evaluating the Econometric Evaluations of Training Programs with Experimental Data." The American Economic Review 76(4): 604-620.

Larson, Andrea (1992). "Network Dyads in Entrepreneurial Settings: A Study of the Governance of Exchange Relationships." Administrative Science Quarterly 37(1): 76-104.

LaTour, Stephen A. and Manrai, Ajay K. (1989). "Interactive Impact of Informational and Normative Influence on Donations." Journal of Marketing Research 26(3): 327-335.

http://www.kickstarter.com/help/faq/kickstarter%20basics.

http://www.kickstarter.com/help/faq/kickstarter%20basics

55 | P a g e

Lin, Ting Hsiang (2010). "A comparison of multiple imputation with EM algorithm and MCMC method for quality of life missing data." Quality & Quantity 44(2): 277-287.

Malone, Thomas, Robert Laubacher and Chrysanthos Dellarocas (2010). "The Collective Intelligence Genome." MIT Sloan Management Review 51(3): 21-31.

McCracken, Harry (2011). The 50 Best Websites of 2011, August 16, 2011. Time.

O'Reilly, Tim (2005). "What Is Web 2.0."

Ordanini, Andrea, Lucia Miceli, Marta Pizzetti and A. Parasuraman (2011). "Crowd-funding: transforming customers into investors through innovative service platforms." Journal of Service Management 22(4): 443-470.

Podolny, Joel M. (1994). "Market Uncertainty and the Social Character of Economic Exchange." Administrative Science Quarterly 39(3): 458-483.

Poetz, Marion and Schreier, Martin (2012). "The Value of Crowdsourcing: Can Users Really Compete with Professionals in Generating New Product Ideas?" Journal of Product Innovation Management 29(2): 245-256.

Reingen, Peter H. (1978). "On Inducing Compliance with Requests." Journal of Consumer Research 5(2): 96-102.

Reynolds, Paul D. (1987). "New firms: Societal contribution versus survival potential." Journal of Business Venturing Volume 2(Issue 3): Pages 231–246.

Rheingold, H. (1993). "The virtual community: Homesteading on the electronic frontier." New York, NY: Harper Collins.

Schibrowsky, John and Peltier, James (1995). "Decision frames and direct marketing offers: a field study in a fundraising context." Journal of Direct Marketing 9(1): 8-16.

Schwienbacher, Armin and Larralde, Benjamin (2010). Crowdfunding of Small Enterpeneureal Ventures. Handbook of Entrepreneurial Finance Oxford University Press: 936

Shane, Scott and Cable, Daniel (2002). "Network Ties, Reputation, and the Financing of New Ventures." Management Science 48 (3 ): 364-381.

56 | P a g e

Smith, Dave (2013). "Pebble Watch Release Date Announced At CES 2013: Why The Original Kickstarter Hit Took So Long To Finally Ship." from http://www.kickstarter.com/blog.

Snyder, Steven James (2010). 50 Websites That Make the Web Great, Nov. 11, 2010. Time, time.com.

Stieger, Daniel, Kurt Matzler, Florian Ladstaetter-Fussenegger and Sayan Chatterjee (2012). "Democratizing Strategy: How Crowdsourcing Can Be Used for Strategy Dialogues." California Management Review 54(4): 44-68.

Trusov, Michel, Anand V. Bodapati and Randolph E. Bucklln (2010). "Determining Influential Users in Internet Social Networks." Journal of Marketing Research 47(4 ): 643-658.

Tukey, J. W. (1977). "Exploratory Data Analysis." Reading, MA: Addison-Wesley.

Turner, John; Oakes, Penny (1986). "The significance of the social identity concept for social psychology with reference to individualism, interactionism and social influence." British Journal of Social Psychology 25 (3): 237–252.

Venkataraman, S. (1997). "The Distinctive Domain of Entrepreneurship Research." Advances in entrepreneurship, firm emergence and growth 3(119-138): 119-138.

Zouhali, Malika (2011 ). Comparison of Crowdfunding Websites. Inc. Magazine.

http://www.kickstarter.com/blog

57 | P a g e

10 Appendix:

Appendix 1

N Minimum Maximum Mean Std. Deviation

Days_Lasted 6289 2 60 32.22 10.51

Pledged 6289 0 4188927 10576.00 81915.06

Goal 6289 1 21474836 28380.81 453713.82

Avr_Pledge 6289 0 2600 63.60 86.23

Success_Ratio 6289 0 179 1.12 4.19

Success 6289 0 1 0.51 0.50

N_Backers 6289 0 74405 153.08 1193.49

N_Updates 6289 0 99 4.03 6.30

N_Comments 6289 0 47827 38.68 861.99

N_Tiers 6289 1 89 9.13 5.82

Tiers_avr 6289 1 6750 372.07 527.55

Tier_range 6289 0 9999 1964.62 2880.31

Tier_max 6289 1 10000 1973.14 2880.24

Tier_min 6289 1 599 8.53 20.26

Previous_success 6287 0 25 0.17 1.08

Previous_failure 6289 0 10 0.13 0.48

Previous_Canceled 6289 0 4 0.04 0.23

Creator_backed 6289 0 524 2.83 9.99

Creator_comments 6289 0 1159 7.47 46.88

Member_since 6289 25 1430 248.60 274.65

FB_friends 4356 0 5247 743.62 864.01

Website 6289 0 1 0.67 0.47

Financial_Plan 6289 0 1 0.03 0.16

Descriptive Statistics

58 | P a g e

Appendix 2

59 | P a g e

Appendix 3

Tra

ns

form

ed

_S

ucce

ss

_R

atio

LN

_G

oa

l

LN

_T

iers

_

Avr

LN

_N

_T

ie

rs

LN

_P

revi

o

us

_S

ucce

ss

LN

_P

revi

o

us

_F

ailu

re

LN

_R

ep

ut

atio

n_

Ind

e

x

LN

_C

rea

t

or_

Ba

cke

d

LN

_C

rea

t

or_

Co

mm

en

ts

LN

_M

em

b

ers

hip

_L

e

ng

th

LN

_F

B_

Fri

en

ds

_M

C

MC

We

bs

ite

Fin

an

cia

l_

Pla

nG

am

es

Film

sM

us

icP

ub

lis

hin

gD

es

ign

Fa

sh

ion

Art

Te

ch

no

log

yT

he

ate

rC

om

ics

Ph

oto

gra

p

hy

Fo

od

Da

nce

Da

ys_

Sin

c

e_

En

d

Da

ys_

La

st

ed

LN

_N

_U

p

da

tes

LN

_N

_C

o

mm

en

ts

Tra

ns

form

ed

_S

ucce

ss

_R

atio

1-,

262*

*-,

043*

*,2

49**

,192

**-.

018

,153

**,3

25**

,409

**,2

01**

,135

**,1

16**

.010

.019

-,02

9*,0

66**

-,09

8**

.006

-,05

1**

.003

-.00

5.0

21,1

18**

-.02

3-.

012

,042

**-,

043*

*-,

131*

*,5

80**

,494

**L

N_

Go

al

-,2

62

**1

,585

**,3

29**

-,09

2**

-,10

7**

.016

,116

**,1

46**

,058

**.0

17,1

56**

-,06

1**

,122

**,0

55**

-,10

3**

-,04

3**

,120

**-.

016

-,13

4**

,152

**-,

041*

*-,

045*

*-,

058*

*,0

89**

-,04

7**

-.00

3,2

10**

,118

**,2

43**

LN

_T

iers

_A

vr-,

04

3**

,58

5**

1,5

11**

-,06

6**

-,09

7**

.021

,073

**.0

00,1

00**

,111

**,1

31**

-,03

7**

-.01

9,1

59**

,026

*-,

095*

*-,

107*

*-,

061*

*-,

036*

*,0

61**

-.00

4-,

053*

*-,

029*

,066

**-.

010

-.00

5,1

05**

,140

**,0

80**

LN

_N

_T

iers

,24

9**

,32

9**

,51

1**

1,0

72**

-.02

0,0

63**

,324

**,2

91**

,200

**,0

77**

,149

**-.

001

,103

**,0

27*

.022

-,13

0**

-.00

9.0

20-,

059*

*-.

001

-,05

3**

,117

**-.

019

,039

**-,

036*

*-,

036*

*,0

45**

,425

**,3

37**

LN

_P

revi

ou

s_

Su

cce

ss

,19

2**

-,0

92

**-,

06

6**

,07

2**

1,2

94**

,563

**,2

67**

,336

**,3

21**

,065

**,0

72**

.004

,038

**-.

004

-.01

6-.

023

.024

-,02

6*.0

02-,

029*

,035

**,0

72**

-,02

6*-,

034*

*,0

29*

.006

-,06

0**

,137

**,1

17**

LN

_P

revi

ou

s_

Fa

ilu

re-.

01

8-,

10

7**

-,0

97

**-.

02

0,2

94

**1

-,47

0**

,114

**,1

92**

,214

**-,

050*

*-.

017

,028

*,0

55**

.015

-,07

4**

-.00

1,0

41**

.023

-.01

9-,

030*

-.01

4,0

60**

.014

-.01

9.0

07,0

32*

-.01

2.0

09.0

24L

N_

Re

pu

tatio

n_

Ind

ex

,15

3**

.01

6.0

21

,06

3**

,56

3**

-,4

70

**1

,111

**,1

13**

,055

**,0

82**

,073

**-.

011

.002

-.01

8,0

51**

-.00

9-.

001

-,04

1**

-.00

5.0

03,0

30*

-.01

2-.

025

-.00

5.0

12-,

029*

-,03

9**

,090

**,0

74**

LN

_C

rea

tor_

Ba

cke

d,3

25

**,1

16

**,0

73

**,3

24

**,2

67

**,1

14

**,1

11

**1

,510

**,4

52**

.009

,162

**.0

24,1

27**

-.02

4-,

119*

*-.

005

,084

**-.

024

-,03

2*,0

28*

-,05

1**

,141

**.0

07.0

12-.

014

-.02

3-.

007

,475

**,3

87**

LN

_C

rea

tor_

Co

mm

en

ts,4

09

**,1

46

**.0

00

,29

1**

,33

6**

,19

2**

,11

3**

,51

0**

1,2

70**

-,03

5**

,148

**-.

004

,320

**-,

098*

*-,

154*

*-,

048*

*,1

83**

-.01

2-,

048*

*,0

99**

-,08

6**

,128

**-,

031*

-.01

4-,

044*

*.0

09.0

22,5

64**

,796

**L

N_

Me

mb

ers

hip

_L

en

gth

,20

1**

,05

8**

,10

0**

,20

0**

,32

1**

,21

4**

,05

5**

,45

2**

,27

0**

1,0

55**

,125

**.0

15,0

50**

.004

-.01

9-.

019

,035

**-,

035*

*-,

027*

.000

.001

,078

**-.

015

-,02

6*.0

15-.

022

.001

,229

**,1

81**

LN

_F

B_

Fri

en

ds

_M

CM

C,1

35

**.0

17

,11

1**

,07

7**

,06

5**

-,0

50

**,0

82

**.0

09

-,0

35

**,0

55

**1

,026

*-,

029*

-,11

7**

,029

*,1

96**

-,06

4**

-,05

5**

.007

-.02

2-,

078*

*,0

31*

-.02

4-.

010

-,05

9**

,047

**-.

017

.016

.022

.010

We

bs

ite

,11

6**

,15

6**

,13

1**

,14

9**

,07

2**

-.0

17

,07

3**

,16

2**

,14

8**

,12

5**

,02

6*

1-.

018

,053

**-,

130*

*.0

14.0

23,0

62**

-.00

6-.

007

,048

**-.

004

.020

.019

-.00

2.0

20-.

014

.011

,185

**,1

57**

Fin

an

cia

l_P

lan

.01

0-,

06

1**

-,0

37

**-.

00

1.0

04

,02

8*

-.0

11

.02

4-.

00

4.0

15

-,0

29

*-.

01

81

-.02

1.0

08-.

006

,032

*-.

023

-.00

5.0

10-.

012

-.01

1,0

29*

.000

-.02

0.0

12.0

05-,

043*

*.0

11-.

023

Ga

me

s.0

19

,12

2**

-.0

19

,10

3**

,03

8**

,05

5**

.00

2,1

27

**,3

20

**,0

50

**-,

11

7**

,05

3**

-.0

21

1-,

144*

*-,

138*

*-,

105*

*-,

068*

*-,

063*

*-,

088*

*-,

049*

*-,

054*

*-,

049*

*-,

044*

*-,

064*

*-,

029*

-.01

9,0

26*

,163

**,3

00**

Film

s-,

02

9*

,05

5**

,15

9**

,02

7*

-.0

04

.01

5-.

01

8-.

02

4-,

09

8**

.00

4,0

29

*-,

13

0**

.00

8-,

14

4**

1-,

263*

*-,

200*

*-,

129*

*-,

120*

*-,

169*

*-,

094*

*-,

102*

*-,

094*

*-,

084*

*-,

122*

*-,

056*

*-.

001

-,03

4**

-.01

9-,

094*

*M

us

ic,0

66

**-,

10

3**

,02

6*

.02

2-.

01

6-,

07

4**

,05

1**

-,1

19

**-,

15

4**

-.0

19

,19

6**

.01

4-.

00

6-,

13

8**

-,2

63

**1

-,19

1**

-,12

4**

-,11

4**

-,16

1**

-,08

9**

-,09

8**

-,08

9**

-,08

0**

-,11

7**

-,05

3**

-.01

3,0

40**

-,03

9**

-,10

6**

Pu

blis

hin

g-,

09

8**

-,0

43

**-,

09

5**

-,1

30

**-.

02

3-.

00

1-.

00

9-.

00

5-,

04

8**

-.0

19

-,0

64

**.0

23

,03

2*

-,1

05

**-,

20

0**

-,1

91

**1

-,09

4**

-,08

7**

-,12

3**

-,06

8**

-,07

4**

-,06

8**

-,06

1**

-,08

9**

-,04

0**

.007

.017

-,08

7**

-,08

6**

De

sig

n.0

06

,12

0**

-,1

07

**-.

00

9.0

24

,04

1**

-.0

01

,08

4**

,18

3**

,03

5**

-,0

55

**,0

62

**-.

02

3-,

06

8**

-,1

29

**-,

12

4**

-,0

94

**1

-,05

6**

-,07

9**

-,04

4**

-,04

8**

-,04

4**

-,04

0**

-,05

7**

-,02

6*.0

09,0

30*

,056

**,1

79**

Fa

sh

ion

-,0

51

**-.

01

6-,

06

1**

.02

0-,

02

6*

.02

3-,

04

1**

-.0

24

-.0

12

-,0

35

**.0

07

-.0

06

-.0

05

-,0

63

**-,

12

0**

-,1

14

**-,

08

7**

-,0

56

**1

-,07

3**

-,04

1**

-,04

5**

-,04

1**

-,03

6**

-,05

3**

-.02

4-.

014

-.01

6-,

035*

*-.

019

Art

.00

3-,

13

4**

-,0

36

**-,

05

9**

.00

2-.

01

9-.

00

5-,

03

2*

-,0

48

**-,

02

7*

-.0

22

-.0

07

.01

0-,

08

8**

-,1

69

**-,

16

1**

-,1

23

**-,

07

9**

-,0

73

**1

-,05

7**

-,06

3**

-,05

7**

-,05

2**

-,07

5**

-,03

4**

.012

-,04

2**

-,04

0**

-,07

2**

Te

ch

no

log

y-.

00

5,1

52

**,0

61

**-.

00

1-,

02

9*

-,0

30

*.0

03

,02

8*

,09

9**

.00

0-,

07

8**

,04

8**

-.0

12

-,0

49

**-,

09

4**

-,0

89

**-,

06

8**

-,0

44

**-,

04

1**

-,0

57

**1

-,03

5**

-,03

2*-,

029*

-,04

2**

-.01

9-.

005

.003

,040

**,1

21**

Th

ea

ter

.02

1-,

04

1**

-.0

04

-,0

53

**,0

35

**-.

01

4,0

30

*-,

05

1**

-,0

86

**.0

01

,03

1*

-.0

04

-.0

11

-,0

54

**-,

10

2**

-,0

98

**-,

07

4**

-,0

48

**-,

04

5**

-,0

63

**-,

03

5**

1-,

035*

*-,

031*

-,04

6**

-.02

1-.

006

-.01

3-,

046*

*-,

075*

*C

om

ics

,11

8**

-,0

45

**-,

05

3**

,11

7**

,07

2**

,06

0**

-.0

12

,14

1**

,12

8**

,07

8**

-.0

24

.02

0,0

29

*-,

04

9**

-,0

94

**-,

08

9**

-,0

68

**-,

04

4**

-,0

41

**-,

05

7**

-,0

32

*-,

03

5**

1-,

029*

-,04

2**

-.01

9-.

005

.013

,137

**,1

08**

Ph

oto

gra

ph

y-.

02

3-,

05

8**

-,0

29

*-.

01

9-,

02

6*

.01

4-.

02

5.0

07

-,0

31

*-.

01

5-.

01

0.0

19

.00

0-,

04

4**

-,0

84

**-,

08

0**

-,0

61

**-,

04

0**

-,0

36

**-,

05

2**

-,0

29

*-,

03

1*

-,0

29

*1

-,03

7**

-.01

7.0

12-,

030*

-,04

1**

-,04

2**

Fo

od

-.0

12

,08

9**

,06

6**

,03

9**

-,0

34

**-.

01

9-.

00

5.0

12

-.0

14

-,0

26

*-,

05

9**

-.0

02

-.0

20

-,0

64

**-,

12

2**

-,1

17

**-,

08

9**

-,0

57

**-,

05

3**

-,0

75

**-,

04

2**

-,0

46

**-,

04

2**

-,0

37

**1

-.02

5,0

28*

.001

.016

.008

Da

nce

,04

2**

-,0

47

**-.

01

0-,

03

6**

,02

9*

.00

7.0

12

-.0

14

-,0

44

**.0

15

,04

7**

.02

0.0

12

-,0

29

*-,

05

6**

-,0

53

**-,

04

0**

-,0

26

*-.

02

4-,

03

4**

-.0

19

-.0

21

-.0

19

-.0

17

-.0

25

1.0

02-.

016

-.01

9-,

040*

*D

ays

_S

ince

_E

nd

-,0

43

**-.

00

3-.

00

5-,

03

6**

.00

6,0

32

*-,

02

9*

-.0

23

.00

9-.

02

2-.

01

7-.

01

4.0

05

-.0

19

-.0

01

-.0

13

.00

7.0

09

-.0

14

.01

2-.

00

5-.

00

6-.

00

5.0

12

,02

8*

.00

21

-.02

0.0

15-.

019

Da

ys_

La

ste

d-,

13

1**

,21

0**

,10

5**

,04

5**

-,0

60

**-.

01

2-,

03

9**

-.0

07

.02

2.0

01

.01

6.0

11

-,0

43

**,0

26

*-,

03

4**

,04

0**

.01

7,0

30

*-.

01

6-,

04

2**

.00

3-.

01

3.0

13

-,0

30

*.0

01

-.0

16

-.0

20

1.0

24,0

34**

LN

_N

_U

pd

ate

s,5

80

**,1

18

**,1

40

**,4

25

**,1

37

**.0

09

,09

0**

,47

5**

,56

4**

,22

9**

.02

2,1

85

**.0

11

,16

3**

-.0

19

-,0

39

**-,

08

7**

,05

6**

-,0

35

**-,

04

0**

,04

0**

-,0

46

**,1

37

**-,

04

1**

.01

6-.

01

9.0

15

.02

41

,583

**L

N_

N_

Co

mm

en

ts,4

94

**,2

43

**,0

80

**,3

37

**,1

17

**.0

24

,07

4**

,38

7**

,79

6**

,18

1**

.01

0,1

57

**-.

02

3,3

00

**-,

09

4**

-,1

06

**-,

08

6**

,17

9**

-.0

19

-,0

72

**,1

21

**-,

07

5**

,10

8**

-,0

42

**.0

08

-,0

40

**-.

01

9,0

34

**,5

83

**1

**. C

orr

ela

tio

n is

sig

nific

an

t a

t th

e 0

.01

le

vel (2

-ta

ile

d).

*. C

orr

ela

tio

n is

sig

nific

an

t a

t th

e 0

.05

le

vel (2

-ta

ile

d).

Co

rre

lati

on

s

60 | P a g e

Appendix 4

Descriptive Statistics (before transformation)

N Minimum Maximum Mean Std.

Deviation Skewness Kurtosis

Statistic Statistic Statistic Statistic Statistic Statistic Std. Error Statistic

Std. Error

Success_Ratio 6118 0.00 9.84 0.84 1.08 3.358 .031 17.547 .063

Goal 6118 50.00 530000.00 14379.55 37150.07 7.682 .031 76.686 .063

Tiers_Avr 6118 2.50 6750.25 370.23 520.29 3.014 .031 14.055 .063

N_Tiers 6118 1.00 59.00 9.05 5.57 2.041 .031 7.731 .063

Reputation_Index 6118 -7.00 19.00 -0.01 0.97 9.878 .031 177.859 .063

Creator_Backed 6118 0.00 75.00 2.49 5.63 5.320 .031 39.358 .063

Creator_Comments 6118 0.00 763.00 5.18 28.77 13.440 .031 243.211 .063

Membership_Length 6118 25.00 1430.00 246.21 272.15 1.666 .031 2.252 .063

FB_Friends_MCMC 6118 16.00 5247.00 820.57 787.85 2.370 .031 7.596 .063

Days_Since_End 6118 0.00 60.00 28.40 17.28 .107 .031 -1.138 .063

Days_Lasted 6118 3.00 60.00 32.21 10.48 .993 .031 1.699 .063

Valid N (listwise) 6118

Values outside range of -1 to +1 for Skewness indicate substantially skewed distribution, Source: Joseph H. Hair (2009)

Descriptive Statistics (after transformation)

N Minimu

m Maximu

m Mean

Std. Deviatio

n Skewness Kurtosis

Statistic Statistic Statistic Statisti

c Statistic Statisti

c Std. Error

Statistic

Std. Error

Transformed_Success_Ratio

6118 0.00 2.38 0.50 0.45 .772 .031 .713 .063

LN_Goal 6118 3.91 13.18 8.46 1.45 .117 .031 .090 .063

LN_Tiers_Avr 6118 0.92 8.82 5.13 1.30 -.034 .031 -.509 .063

LN_N_Tiers 6118 0.00 4.08 2.03 0.61 -.537 .031 1.151 .063

LN_Reputation_Index 6118 0.00 3.26 1.94 0.12 -1.176 .031 66.025 .063

LN_Creator_Backed 6118 0.00 4.33 0.73 0.88 1.142 .031 .707 .063

LN_N_Comments 6118 0.00 8.33 0.81 1.24 1.933 .031 4.057 .063

LN_Membership_Length

6118 3.22 7.27 4.96 1.04 .375 .031 -1.133 .063

LN_FB_Friends_MCMC

6118 2.83 8.57 6.30 0.98 -.508 .031 .352 .063

LN_Days_Since_End 6118 0.00 4.11 3.11 0.87 -1.175 .031 .880 .063

LN_Days_Lasted 6118 1.39 4.11 3.45 0.33 -.918 .031 4.219 .063

Valid N (listwise) 6118

Values outside range of -1 to +1 for Skewness indicate substantially skewed distribution, Source: Joseph H. Hair (2009)

61 | P a g e

Appendix 5

62 | P a g e

63 | P a g e

64 | P a g e

Appendix 6

65 | P a g e

Appendix 7

Standardized

Coefficients

B Std. Error Beta

(Constant) .185 .109 1.698 .090

LN_Days_Since_End -.008 .005 -.015 -1.437 .151

LN_Days_Lasted -.077 .015 -.056 -5.109 .000

LN_N_Tiers .185 .010 .252 18.724 .000

Games .038 .022 .022 1.758 .079

Music .022 .015 .020 1.470 .142

Publishing -.081 .017 -.060 -4.691 .000

Design .077 .024 .040 3.288 .001

Fashion -.099 .024 -.048 -4.112 .000

Art -.036 .019 -.023 -1.878 .060

Technology .138 .030 .053 4.672 .000

Theater .055 .027 .023 2.027 .043

Comics .100 .030 .038 3.339 .001

Photography -.116 .032 -.040 -3.623 .000

Food .039 .024 .019 1.652 .099

Dance .116 .047 .027 2.480 .013

LN_FB_Friends_MCMC .049 .005 .107 9.706 .000

LN_Membership_Length .022 .007 .050 3.284 .001

LN_Creator_Backed .168 .030 .329 5.578 .000

LN_Goal -.128 .004 -.411 -28.419 .000

LN_Tiers_Avr .007 .005 .020 1.274 .203

External_Platform .078 .010 .082 7.428 .000

Financial_Plan -.040 .030 -.014 -1.333 .182

LN_Reputation_Index .368 .042 .095 8.755 .000

Membership_Length_Cre

ator_Backed

-.008 .005 -.100 -1.581 .114

F 117.73

df 6117.00

R Square 0.32

Adjusted R Square 0.31

Model Summary

Interaction Model

(Membership and

Involvement)

Unstandardized Coefficients

t Sig.

66 | P a g e

Appendix 8

B S.E. Wald df Sig. Exp(B)

LN_Days_Since_End -.085 .035 6.017 1 .014 .919

LN_Days_Lasted -.555 .097 32.788 1 .000 .574

LN_N_Tiers 1.003 .067 221.760 1 .000 2.726

Games -.516 .138 13.954 1 .000 .597

Music .315 .094 11.270 1 .001 1.371

Publishing -.691 .110 39.188 1 .000 .501

Design -.225 .149 2.280 1 .131 .798

Fashion -.954 .153 39.013 1 .000 .385

Art -.272 .116 5.505 1 .019 .762

Technology -.010 .191 .003 1 .958 .990

Theater .722 .171 17.886 1 .000 2.059

Comics -.191 .198 .935 1 .334 .826

Photography -1.012 .201 25.436 1 .000 .364

Food .087 .145 .365 1 .546 1.091

Dance 1.139 .332 11.790 1 .001 3.124

LN_FB_Friends_MCMC .348 .033 110.534 1 .000 1.416

LN_Membership_Length .059 .032 3.328 1 .068 1.061

LN_Creator_Backed .708 .043 265.216 1 .000 2.030

LN_Goal -.692 .033 452.547 1 .000 .501

LN_Tiers_Avr .114 .035 10.837 1 .001 1.121

External_Platform .432 .066 42.741 1 .000 1.541

Financial_Plan .031 .195 .026 1 .872 1.032

Constant 2.283 .436 27.370 1 .000 9.806

Cox & Snell R Square .265

Nagelkerke R Square .353

% Correct Prediction 73.1

Model Summary

Binary Logistic Regression Model

Dependent Variable = Project Success

67 | P a g e

Appendix 9

Beta p Beta p Beta p Beta p Beta p

(Constant) .000 .000 .000 .000 .000

LN_Days_Since_End -.017 .103 -.018 .096 -.017 .103 -.017 .110 -.018 .100

LN_Days_Lasted -.060 .000 -.060 .000 -.060 .000 -.060 .000 -.061 .000

C_LN_N_Tiers .257 .000 .248 .000 .257 .000 .258 .000 .255 .000

Games .021 .086 -.066 .152 .022 .728 -.008 .638 .224 .001

Music .025 .070 .025 .071 .057 .485 .042 .008 .026 .052

Publishing -.059 .000 -.066 .044 -.059 .000 -.059 .000 -.058 .000

Design .040 .001 .040 .001 .040 .001 .041 .001 .039 .002

Fashion -.050 .000 -.050 .000 -.050 .000 -.050 .000 -.050 .000

Art -.022 .076 -.022 .069 -.022 .078 -.022 .078 -.001 .985

Technology .055 .000 .055 .000 .055 .000 .055 .000 .091 .146

Theater .026 .020 .026 .023 .026 .020 .027 .020 .027 .017

Comics .036 .002 -.029 .554 .036 .002 .041 .023 .037 .001

Photography -.042 .000 -.042 .000 -.042 .000 -.042 .000 -.041 .000

Food .020 .092 .020 .087 .020 .090 .020 .089 .019 .105

Dance .028 .009 .028 .010 .028 .009 .028 .009 .029 .008

C_LN_FB_Friends_MCMC .113 .000 .113 .000 .115 .000 .114 .000 .114 .000

LN_Membership_Length .035 .004 .035 .003 .035 .004 .037 .002 .034 .005

C_LN_Creator_Backed .248 .000 .246 .000 .248 .000 .247 .000 .249 .000

C_LN_Goal -.409 .000 -.409 .000 -.409 .000 -.410 .000 -.396 .000

LN_Tiers_Avr .018 .253 .018 .250 .018 .251 .018 .242 .017 .273

External_Platform .087 .000 .087 .000 .087 .000 .087 .000 .086 .000

Financial_Plan -.015 .148 -.015 .156 -.015 .150 -.015 .160 -.015 .150

C_LN_N_Tiers_Comics .068 .169

C_LN_N_Tiers_Games .091 .049

C_LN_N_Tiers_Publishing .007 .832

C_LN_FB_Friends_MCMC_Games -.001 .988

C_LN_FB_Friends_MCMC_Music -.033 .688

C_LN_Creator_Backed_Games .041 .010

C_LN_Creator_Backed_Comics -.007 .713

C_LN_Goal_Games -.207 .002

C_LN_Goal_Art -.019 .764

C_LN_Goal_Technology -.039 .538

Adjusted R Square .306 .306 .305 .307 .306

F 123.428 108.883 113.115 109.358 109.104

Project Categories InteractionsBase Int. Int. Int. Int.

68 | P a g e

Appendix 10

69 | P a g e

Appendix 11

social crowdfunding: individual- and project-related ... · kind of fundraising, websites with...

Documents